Book an Assessment
Question Library


82 QUESTIONS LEADERS ASK ABOUT AI, ANSWERED PLAINLY.

Each answer is short first, then detailed, with next steps and the sources behind it. Drawn from Your AI Wake-Up Call by Wade W. Wyant, and cross-referenced to 35 outside sources you can open and check. 82 questions · 35 cited outside sources 

All 82 questions

Grouped by the work you need done: Training & Enablement first, then Design & Strategy, then Implementation. Each answer says which offering it belongs to. Select a question to open the answer.
How do we train staff to use AI?
Short answer. Treat it as ongoing enablement, not a one-time training event. Give people a real foundation in how the technology works, put a human coach beside them for the first hours, and keep that support going as the tools change. Wade Wyant's core point in the final chapter of Your AI Wake-Up Call is to stop calling it training. You take driver's training once and almost never go back, because driving stays learned. AI keeps changing and will be with your people for the rest of their careers, so a single session with a certificate at the end does not hold.

Most of the early value comes from foundations, not prompt tricks. The book compares it to Microsoft Office: you do not need to know how a computer works, but you do need to be good at Word, Excel and Outlook. Your people need enough understanding of the technology to look at their own job and see where it fits.

The part that surprises most leaders is the delivery. People do not want an AI teaching them AI. They want a person who knows the tools to sit with them for a couple of hours, watch them work through a real task and reassure them when they feel lost. The book reports that the return on those hours shows up in confidence as much as adoption, and confident users find uses nobody designed for. What to do next
  1. Replace the one-time kickoff with a written, ongoing enablement plan with named human touchpoints.
  2. Start with the foundation: what the tools are, where they fail, and how they connect to your data and other tools.
  3. Schedule short, hands-on sessions where a coach works on each person's real tasks.
  4. Revisit the plan when tools change, when new people join, and when the technology shifts.
 Offering: Training & Enablement
Relevant to: CEO, HR and people, Owner-operator

Sources
Why call it AI enablement instead of AI training?
Short answer. Training implies a finished event. AI is a pervasive technology that keeps changing, so the job is ongoing enablement: coaching, connections to real work, and human support over time. The book frames it as one of its biggest mindset shifts. Training ends with a certificate. Enablement does not end, because the tools, the models and your own use cases keep moving.

In 2026, the book argues, enablement mostly stops meaning how to write a good prompt. It means someone sitting with you to make a connection you would otherwise need an IT professional for, helping you build a first agent, or walking you through building something with vibe coding. Prompting skill still helps, and it is worth having someone on the team who is good at it, but it is no longer the main event.

Wyant also notes that the label matters less than the outcome. Call it training, enablement or coaching. What counts is people who use AI well, trust it, and keep using it. What to do next
  1. Rename the program internally so nobody expects a finish line.
  2. Budget for coaching hours each quarter, not a single workshop.
  3. Measure confidence and continued use, not attendance.
 Offering: Training & Enablement
Relevant to: CEO, HR and people

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 56
Is prompt engineering the main skill our employees need?
Short answer. No. Prompting matters, but the real skill is understanding the technology well enough to see where it fits your own job. A prompt cheat sheet is not a foundation. Chapter 51 lists enablement as the first place AI rollouts fall apart, and says most people misread what enablement means. Knowing how to write a good prompt matters, but that is not the real skill.

The book's analogy is office software. You can be productive in Microsoft Office without understanding how a computer works, but you need a decent handle on Word and Excel. Advanced AI tools are no different. People need a real base understanding so they can map the tool onto their own work.

The final chapter adds that having one person on the team who is strong at prompting is useful for now, but that the larger value is in connecting tools, building first agents, and learning to build small tools with AI. What to do next
  1. Teach how the tools work, where they fail, and how to check output before teaching prompt patterns.
  2. Have each person bring one real task to the first session.
  3. Name one prompting specialist as a resource, not as the whole program.
 Offering: Training & Enablement
Relevant to: CEO, HR and people, IT and technology

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 51; Ch. 56
Why do employees learn AI better from a person than from an AI tutor?
Short answer. Research and the author's experience agree that people want another human to support them through learning AI. A person who knows the tool, watches you work and reassures you builds confidence that self-serve material does not. The book points out the irony. AI can teach you to configure a sprinkler system or write code, and Wyant used it that way himself. Yet when the subject is AI itself, people prefer a human.

The model he describes is simple: a person sits with another person for a couple of hours, walks through the tool, answers questions and offers encouragement. He compares it to the Genius Bar. Even the most intuitive product in the world, Apple's iPhone, ships with a human help desk because at some point people need one.

The same chapter adds a coaching analogy. Lionel Messi has a coach. The greatest in any field lean on another set of eyes, and the book asks why an AI rollout should be any different. What to do next
  1. Pair each employee with a coach for their first live session.
  2. Make the first session about a task they do every week.
  3. Keep a regular office-hours slot so help stays available after the rollout.
 Offering: Training & Enablement
Relevant to: CEO, HR and people

Sources
How do we find out where each employee stands on AI?
Short answer. Not with a survey. Create a safe environment, bring in someone with real experience, and give people permission to learn at their own pace. Attitudes range from fear to eager experimentation, and you probably do not know where each person sits. Chapter 56 warns that leaders rarely know where their team stands, and the range is wider than expected. Some people believe AI is an existential threat and have never said so. Others have used ChatGPT on their phones for two years and just need better tools and direction. In between are the executive afraid of looking incompetent, the curious but cautious manager, and the high performer privately worried their expertise will feel obsolete.

The book's answer is that you learn where people stand through a safe, supported learning environment rather than a questionnaire, because people will not say the honest thing on a form. What to do next
  1. Run small group sessions led by an experienced coach, not by the manager.
  2. Tell people explicitly that learning pace is theirs to set.
  3. Listen for fear and eagerness, and plan different support for each.
 Offering: Training & Enablement
Relevant to: CEO, HR and people

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 56
How do we handle employees who are afraid AI will take their jobs?
Short answer. Name the fear directly and explain what AI changes. The book argues AI takes analytical weight off people and leaves judgment, empathy, vision and relationships, and that the realistic threat to a job is a person who uses AI well, not the AI itself. Wyant is direct that the fear is not irrational. Every technology that expanded human capability changed what people were expected to do. He also reports that, so far, jobs people expected to disappear, such as coding, are busier than ever, because the human need for more grows as AI adds capacity.

His line is that people should not worry about AI taking their job, they should worry about a human with AI taking it. That is an argument for enablement: the people who are taught to use the tools well are the ones protected.

Chapter 56 adds that resistance often has nothing to do with the technology. People resist what technology implies about their value. Chapter 23 describes the pattern: people automate pain, tolerate ease, fight difficulty and protect identity. What to do next
  1. Say plainly which tasks AI will take on and which human work you want more of.
  2. Offer coaching first to the people most worried, not last.
  3. Do not roll out tools without explanation. The book describes people who lost jobs to a tool nobody explained.
 Offering: Training & Enablement
Relevant to: CEO, HR and people

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Intro; Ch. 5; Ch. 23; Ch. 47; Ch. 56
Why won't our people use the AI tools we already bought?
Short answer. Usually because the basics were never checked. Chapter 51 asks whether real enterprise accounts exist, whether people use them, whether they can collaborate inside them, and whether they are connected to the data and tools they need. The book says this is almost without exception the first place IT drops the ball: not because IT failed at plumbing, but because nobody verified the rollout end to end. Do you have business accounts rather than personal logins? Are people actually using them, or did IT roll them out and walk away? Are people trained to collaborate inside the tools, not just alone at their desks? Are the tools connected to the data and systems people use every day?

Wyant also warns that the ease of getting started is part of the trap. You sign up, you log in, it works more or less out of the box, and installation gets mistaken for success. Getting the tool installed is not the same as getting it right, and the book says most companies are doing about one percent of one percent of what the technology can do. What to do next
  1. Audit the five questions above for every tool you license.
  2. Find who actually logs in each week and ask them what they use it for.
  3. Connect the tools to the data people need, then retrain on real tasks.
 Offering: Training & Enablement
Relevant to: CEO, IT and technology, HR and people

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 51
What does an ongoing AI enablement plan look like?
Short answer. A long-term plan with human touchpoints at every stage: during implementation, as tools change, when new people join, and when the technology shifts in ways nobody predicted. Not a kickoff and not a single session. The book's closing advice is to build an enablement plan, specifically not a kickoff and not a one-time training session. It should cover how the organization develops AI capability over time.

Wyant observes that the companies winning with AI are not only those that deployed the best platform. They deployed it and kept a human in the loop, starting with enablement or weaving it in from the beginning, and they have not stopped. He summarizes the pairing as strong private AI infrastructure plus genuine human coaching around it.

He also notes that by 2027 enablement is becoming an industry-wide theme alongside data, and puts it plainly: people are not a variable in your AI strategy, they are the strategy. What to do next
  1. List the four trigger points: launch, tool change, new hire, technology shift.
  2. Assign a human owner and a coach for each trigger.
  3. Review the plan every quarter alongside your governance review.
 Offering: Training & Enablement
Relevant to: CEO, HR and people, Owner-operator

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 56
How do we check AI output before it goes to a client or the board?
Short answer. Never treat an AI-generated fact as final until a person has checked it, especially in a board deck, contract or anything a customer will read. AI's confidence has no relationship to its accuracy. Chapter 10 gives three principles. First, check every fact before it ships, because a wrong answer sounds exactly as convincing as a right one. Second, recognize that generic AI is guessing at your industry, market and numbers, while private AI draws on your actual data, but that you still verify rather than assume. Third, put checkpoints in the workflow anyway: any output driving a real decision gets a human's eyes before the decision is made.

The goal, in the book's words, is not betting on a perfect streak. It is making sure nobody ships a hallucination nobody caught.

Real examples in the book show the cost. A New York lawyer filed a brief citing six invented cases. Zillow's algorithmic home buying lost about $569 million after acting on confident numbers no human checked. What to do next
  1. Define which outputs require human review: board materials, contracts, customer-facing text, financial figures.
  2. Require sources for any factual claim and check them.
  3. Make checking a visible norm, so people who skip it are called out by peers.
 Offering: Training & Enablement
Relevant to: CEO, CFO and finance, Legal and compliance

Sources
What is an AI hallucination and how should staff handle it?
Short answer. A hallucination is when AI states something untrue or cites a source that does not exist, with full confidence. It happens because the model predicts plausible text and does not stop to say it does not know. Staff should verify before relying on it. The book describes it with the unicorn-or-dragon question: executives argue passionately about which fictional creature is more real, and AI was trained on exactly that kind of disagreement. It did not learn a clean answer, it learned a contested one. AI does not lie, it guesses confidently.

Wyant cites a 2025 study in which an AI model wrote literature reviews and nearly two thirds of its citations were fabricated or wrong, with the rate worsening on unfamiliar topics. His rule: the better it sounds, the more you check it.

He also says hallucination rates have fallen but not disappeared, and that the rarer it gets, the more it surprises people, which is why guardrails should stay in place rather than being loosened. What to do next
  1. Teach staff the mechanism so they treat fluent output as a draft.
  2. Ask for sources and open them.
  3. Report hallucinations in a shared place so the team learns where the tools fail.
 Offering: Training & Enablement
Relevant to: CEO, Legal and compliance, IT and technology

Sources
How do we stop an analytical executive from checking three AI models before every decision?
Short answer. Recognize it as paralysis dressed up as oversight. The goal is to check AI enough that you can trust your own judgment again, not to check it forever. Move from one-on-one model use toward shared, collective AI use. Chapter 11 describes a private banker CEO who ran three different models in a single meeting and checked everything against all three before committing. The book calls this the data hole: AI makes one more angle, one more model's opinion and one more data cut effortless, so analytical people keep going until the volume of information becomes the problem.

Wyant's view is that both extremes are the same failure to trust your own judgment. Overcaution is not the opposite of recklessness. He admits he sometimes checks one model against another before trusting his own gut, and says the aim is to check enough that you can trust yourself.

He also pushes back on the quote people attribute to Deming, 'In God we trust, all others bring data,' noting the attribution is incorrect and that someone who accepts only data is as stuck as someone who refuses to look at it. What to do next
  1. Agree in advance on which decisions need a second model and which do not.
  2. Give analytical leaders a shared source of truth so they are not re-litigating inputs.
  3. Set a time limit for verification on routine decisions.
 Offering: Training & Enablement
Relevant to: CFO and finance, CEO, Financial services

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 11; Ch. 12
How do we use AI as a thought partner in meetings, not just a search tool?
Short answer. Put it in the room while you decide, and let it push back. The book describes three stages: answer machine, thought partner, then automation. Most leaders skip the middle stage, which matters most. Chapter 52 describes stage one as asking a question and getting an answer, and stage three as letting AI act on its own. Stage two, the one almost nobody talks about, is having AI live in the room while you make a decision. Smart leaders who use AI constantly for research and drafting still do their real thinking the old way, in the moment.

Chapter 53 gives an example. During an annual planning meeting, the team ran its thinking through an early thought-partner tool and asked it to critique them. It argued the company could reach four times its size in two years rather than the five-year goal they had set. The company is on track to hit that mark about three years ahead of the original plan. Wyant is careful to say the technology did not do the work: it gave the room conviction to aim at the bigger number, and the people had to listen.

He is honest that he does not fully understand why leaders resist. People take feedback from a computer more easily than from a colleague, but many still will not let it challenge them in front of their team. What to do next
  1. Pick one recurring meeting, such as quarterly planning, and bring AI into it.
  2. Ask it explicitly to critique the plan and name assumptions.
  3. Record the session so the discussion becomes data for later.
 Offering: Training & Enablement
Relevant to: CEO, COO and operations, Owner-operator

Sources
What is a Claude.md file or personal AI profile, and why use one?
Short answer. A short standing brief about who you are, what your company does, how you think and what matters most, pasted or loaded at the start of every AI session. It is the fastest free upgrade a leader can make today. The book says any leader can do this without a developer. Because AI tools carry little memory of you between sessions, you write your core context once and give it to the AI at the start. Some people call it a Claude.md file, others a personal AI profile. They are the same thing.

The author compares it to briefing a consultant from scratch at every meeting versus working with one who already knows your company and priorities. It is a starting point, not a replacement for a full private AI build, and it makes a later build better because you have already practiced articulating what the machine needs to know.

In Chapter 45 the same document is described in a deeper form: the CEO Doctrinal Statement, written first and then given to the AI. Writing it is step one, giving it to the AI is step two. What to do next
  1. Write one to two pages: who you are, what you do, how you decide, what you never compromise on.
  2. Load it at the start of each session or store it where your tools can read it.
  3. Update it quarterly.
 Offering: Training & Enablement
Relevant to: CEO, Owner-operator

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 8; Ch. 45
Why should the CEO use AI first?
Short answer. Because AI handed to IT, marketing or a committee never aligns with strategy. The book's rule is that the constraint at the top limits everything below it, so AI starts in the office where decisions are made. Chapter 20 says that in almost every implementation the team was brought in to fix, leadership had handed AI to IT, marketing or a committee, and then wondered why it never scaled or aligned with strategy. You cannot start AI at the bottom of the organization and expect transformation at the top.

Chapter 25 adds that the people best equipped to lead implementation are business analysts and strategists, not coders, because the technical side is easy and teaching the system how you think is the hard part.

The book also notes that the intelligence of your system will only match the intelligence of your leadership. If leadership rhythms, meetings and decision reviews are weak, the AI inherits that weakness. If they are strong, it amplifies them. What to do next
  1. Have the CEO use AI personally on real decisions for 30 days before any company-wide rollout.
  2. Name a business owner for the initiative, not only a technical owner.
  3. Move to the leadership team next, then the wider company.
 Offering: Training & Enablement
Relevant to: CEO, Board and investors

Sources
How do we build a culture where people share what works with AI?
Short answer. Keep the human spark alive. AI spread person to person, not through campaigns, and your adoption will spread the same way if you celebrate breakthroughs and let people talk about them. The book notes AI did not spread through marketing or press releases. It spread because someone at lunch told someone else it saved them three hours. That human network effect is still the most powerful force in adoption.

Wyant's advice is not to let AI feel imposed from the top down. Build a culture where sharing what works, what surprised you and what you discovered is celebrated. He calls that culture a compounding advantage: the tool gets better, the people get better, and the gap to competitors who skipped the human layer widens each quarter. What to do next
  1. Create a standing slot in team meetings for one AI discovery.
  2. Recognize people who share, not only people who produce.
  3. Capture shared wins in a common place so they become reusable.
 Offering: Training & Enablement
Relevant to: CEO, HR and people

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 56
How should a CFO use AI?
Short answer. As a weighted voice inside a shared company brain, with a human checkpoint on anything touching financial reporting. The book suggests the CFO's cost insight should carry more weight on financial decisions, and that AI can produce a daily risk profile without a manual report. Chapter 36 gives the example of weighting: the CFO's insight on cost counts more on financial decisions, the CMO's input counts more on brand decisions, and the CEO's strategic intent runs across all of it. Chapter 49 describes automation freeing the CFO from running reports to get a daily, AI-curated risk profile.

On the control side, Chapter 27 says SOX asks whether AI touches anything that feeds financial reporting. A tool that summarizes meetings is a different risk from one that automates journal entries or flags exceptions in financial controls. Where AI touches that pipeline, internal audit must be able to explain what the model does, test it for drift and error like any other control, and keep a human sign-off where SOX has always required one.

Chapter 11 also warns against the data hole: a data-loving executive can use AI to go looking for one more answer indefinitely. What to do next
  1. Separate AI uses that touch financial reporting from those that do not.
  2. Document human sign-offs for any AI-assisted control.
  3. Use a shared AI environment so finance, sales and operations work from the same numbers.
 Offering: Training & Enablement
Relevant to: CFO and finance, Financial services

Sources
How should a sales leader use AI with call transcripts and KPIs?
Short answer. Feed transcripts and performance data into a shared system over time, so the AI can form a historical view of reps and deals instead of a one-time snapshot. The book cites a client with about 1,000 transcripts over nine months. Chapter 14 describes a customer with roughly nine months of transcripts, about a thousand, plus KPIs from sales meetings. Because the AI worked with a historical record rather than a flat import, it could pick out more nuance about a strong sales leader and give sharper input on what that leader could do next.

Chapter 21 notes the same pattern: a customer signal buried in a thousand sales call transcripts that no human had time to read can surface once AI reads them all. Chapter 36 adds that the AI should weight inputs by role and experience, so a veteran seller's read counts more than a new hire's guess.

Chapter 23 adds that research suggests people rate each other less objectively than they believe, which is an argument for letting AI help evaluate performance, with clean data and human review. What to do next
  1. Record and transcribe sales meetings consistently.
  2. Tag speakers so the AI knows who is talking.
  3. Combine transcripts with KPI data and review the AI's read against reality quarterly.
 Offering: Training & Enablement
Relevant to: Sales leader

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 14; Ch. 21; Ch. 23; Ch. 36
  • Nine Lies About Work, Marcus Buckingham and Ashley Goodall, Harvard Business Review Press, 2019
What does AI governance look like for a 50-person firm?
Short answer. A leadership decision set, not an IT project: what information gets preserved, which tools are sanctioned for which uses, how AI output connects to documented decisions, and who is accountable. Then review it every quarter. The book frames governance in Chapter 18 as a leadership strategy problem that happens to have technical parts. It requires executive decisions on four things: what information gets preserved and how, what tools are sanctioned for which uses, how AI output connects to documented decisions, and who owns that documentation. IT can support implementation but cannot own the strategy.

Size matters less than it sounds. Most employees open a browser tab, ask a question, use the answer and close the tab, so there is no record of the prompt, the response or how it influenced the call. The book contrasts that with email, which has thirty years of retention policy and discovery practice behind it. AI has no equivalent norm yet.

Chapter 18 also stresses maintenance. An AI strategy is not evergreen. The book describes companies aligning in January and treating it as finished by March, with the regulatory environment and tool list changed by September. Colorado's AI law was signed in 2024, delayed in 2025 and revised in May 2026. Plan to revisit quarterly.

Applying this to a firm of 50 is our reading of the book's framework: it is small enough that one leadership team can own all four decisions directly. What to do next
  1. List every AI tool in use, sanctioned or not.
  2. Decide which tools are approved for which kinds of data and decisions.
  3. Define how AI-assisted decisions are documented and who signs off.
  4. Put a quarterly governance review on the calendar.
 Offering: Design & Strategy
Relevant to: CEO, Owner-operator, Legal and compliance, IT and technology, Professional services

Sources
Who should own AI strategy: IT or the CEO?
Short answer. The CEO and leadership team own strategy. IT owns the plumbing, security and guardrails. The book says AI is a leadership initiative wearing a technology costume, and that handing it to IT hides the problem somewhere nobody is looking. Chapter 17's line is that you cannot run AI like an IT project, you run it like a leadership initiative. The decisions that matter, such as which voices, biases and values belong in the system, are not technical. Chapter 18 uses a neurosurgeon-versus-therapist analogy: the brain is involved, but the nature of the problem is not surgical.

The book also marks where IT is the right owner. Chapter 51 calls getting the plumbing right the single thing IT should be doing: enterprise accounts, access, connections to data and tools. Chapter 50 says that when automation can touch your systems, move money or change records, securing it is IT's lane and they should be brought in.

A related observation: most companies now roll out licenses to the whole staff, call it a strategy, and leadership never teaches the system how it thinks. Chapter 20 notes the same failure in nearly every project the team was called in to fix. What to do next
  1. Name a business owner for AI, at the executive level.
  2. Give IT explicit ownership of accounts, security, access and integrations.
  3. Keep strategy decisions, such as use cases and data priorities, with leadership.
 Offering: Design & Strategy
Relevant to: CEO, IT and technology, Board and investors

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 17; Ch. 18; Ch. 20; Ch. 50; Ch. 51
Why do most AI initiatives fail?
Short answer. The book argues the failure is rarely technical. It comes from leadership: no owner for data decisions, uncleaned data, treating AI as an IT project, no weighting, no testing, and trying to build everything at once. Chapter 12 cites a 2025 MIT report, as covered by Forbes, saying about 95 percent of corporate AI initiatives show zero return, and that of the companies that ran a real integrated pilot, roughly one in four worked. Wyant adds that his team has been called in to rescue projects with millions already spent, and the technology worked fine.

The appendix lists where builds fail, in order of how often he sees it: nobody owns the decision about what data matters; the data never gets cleaned; it is treated as an IT project; nobody weights the data; testing gets cut when deadlines tighten; and people try to build the whole thing at once instead of starting with one real problem. He calls every one a leadership failure.

He also notes that technology does not fix misalignment, it exposes it, and that failure often follows leaders who adopt AI blindly or hand off accountability along with responsibility. What to do next
  1. Assign an owner for data decisions before you buy anything.
  2. Start with one use case and prove it.
  3. Reserve time and budget for data cleanup and testing.
 Offering: Design & Strategy
Relevant to: CEO, CFO and finance, Board and investors

Sources
Should our AI goal be efficiency or innovation?
Short answer. Both, deliberately. Efficiency is the smaller goal: it makes you faster on the road you are on. The book argues that innovation, finding things you could not see before, is what creates a different business model, and that the question has to be asked on purpose. Chapter 1 argues most CEOs say the word efficiency when asked what they want from AI. It has measurable ROI and is easy to justify, but it is the equivalent of a faster horse. Efficiency optimizes the road you are on, innovation finds a road your competitors do not know exists.

Innovation in the book's sense means using AI to find new markets you are positioned to enter, product extensions your customer data has suggested for years, pricing inefficiencies, patterns that predict churn six months ahead, and talent risk buried in HR data. None of these come from your current processes.

The practical point: tell a team the goal is to save time and they will save time. They will not spontaneously ask AI for the business model you are missing. That question has to be built into the workflow and the doctrine you give the AI. Wyant suggests starting with efficiency if you need quick wins to justify the investment. What to do next
  1. Ask leadership: what is the most important market insight we are missing?
  2. Build that question into a recurring AI review.
  3. Track both efficiency wins and new insights surfaced.
 Offering: Design & Strategy
Relevant to: CEO, Board and investors

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 1
What is a co-CxO and what does it do?
Short answer. A private AI trained on your company's data and leadership thinking, acting as a thought partner at the C-suite level. It shares your worldview and priorities, challenges assumptions and keeps you accountable. It supports leaders and does not replace them. The book uses co-CxO for AI that is a partner in function and not in title. Trained on your values, mission and strategy, it stops being a generic answer machine. An answering machine reacts, a co-CxO thinks with you.

Wyant lists what it delivers in five ways: reclaimed hours, fewer undetected blind spots, scale without starting over, positioning ahead of competitors, and an intelligence asset that compounds. Chapter 9 describes the MIRATECH case, where revenue grew 450 percent in under three years and the intelligence layer already carried the company's strategy and history, so a small team was not starting from scratch each time growth forced a new call.

The book is also explicit that it is not a perfect clone and not a replacement. The point is to amplify judgment and carry the organization's thinking when the leader is not in the room. What to do next
  1. Start with the leadership team's strategy, values and decision history.
  2. Give it permission to challenge, not just summarize.
  3. Build it in phases and expect compounding value after several quarters.
 Offering: Design & Strategy
Relevant to: CEO, Board and investors

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Intro; Ch. 9; Ch. 26
What is a shared company brain, and why does it matter?
Short answer. One AI environment, shared by the whole leadership team and fed by everyone's data, instead of a dozen private tools that never connect. The book calls the individual version a second brain and argues companies need a shared one. Chapter 28 describes the problem. Every leader has a line to their own AI: the CEO to ChatGPT, the CFO to Claude, the COO to Gemini. Each gets sharper, but the lines never connect, so the team walks into meetings with different versions of reality, each reinforced by a machine that told them they were right. In the book's phrase, AI privatized conviction.

The fix is architectural: all those lines feed one shared system, so the CFO asking about financial risk gets an answer informed by the CEO's strategic context, the sales pipeline and operational constraints, and the same answer reaches everyone else. The book says this is alignment, not groupthink: the conflicts that remain are real ones.

The terms shared company brain, sandbox and 3D AI mean the same thing in the book. A 2026 survey cited in Chapter 29 of 1,200 C-suite executives found 54 percent said adopting AI is tearing their company apart. What to do next
  1. Audit which AI tools each leader uses and with what data.
  2. Stand up one shared environment with common context and permissions.
  3. Then widen to frontline managers and their data over time.
 Offering: Design & Strategy
Relevant to: CEO, COO and operations, CFO and finance

Sources
What is the Decision Alignment Layer?
Short answer. The layer where your values, risk tolerance and strategy shape how AI informs decisions, and where AI's role in decisions is tracked. The book says most AI efforts fight over models and infrastructure while this layer stays wide open. Chapter 48 maps the stack as Gartner describes it: infrastructure and models at the bottom, with decisions, intelligence and execution above. Most of the industry competes at the bottom, and the layers where business value lives are open. The book defines the Decision Alignment Layer as the part that holds your strategy, values and risk tolerance, and cites Gartner's projection that by 2027 half of business decisions will be augmented or automated by AI.

In Chapter 18 it is also the governance function: tracking decisions, documenting AI's role and building the audit trail that state laws already demand. Chapter 29 ties the term to the privatized-conviction problem, solved by one shared brain.

REDEGADES.AI builds this layer for clients. This page describes the concept as the book frames it. What to do next
  1. Write down the values and risk limits AI must respect.
  2. Capture decisions and the role AI played in each.
  3. Put that record in the shared environment, not in individual chats.
 Offering: Design & Strategy
Relevant to: CEO, COO and operations, Board and investors

Sources
Why do our leaders disagree more since we adopted AI?
Short answer. Often it is false disagreement: two people arguing from different AI outputs and different inputs, not a real strategic conflict. The book's fix is one shared brain, so the disagreements that remain are real. Chapter 14 describes leaders who each run their own AI, walking into the most expensive meeting more convinced than ever. They have data and analysis, and an argument that sounds airtight because a machine built it. Wyant says you did not build a smarter company, you built a company of smarter individuals who agree less.

He names three costs: time (the drama tax of relitigating settled decisions), trust (people stop believing decisions are real) and momentum (stalled decisions stall execution). The false disagreement problem is that you can no longer tell a real fight from a data mismatch. Fix the inputs and the fight often disappears.

The aim is not a room that never disagrees. Disagreement is healthy. False disagreement is expensive. What to do next
  1. Ask whether the disagreement survives when both people use the same inputs.
  2. Move leaders onto a shared AI environment and shared data.
  3. Track how many decisions get reopened each quarter.
 Offering: Design & Strategy
Relevant to: CEO, COO and operations, CFO and finance

Sources
How do we get AI to challenge us instead of just agreeing with us?
Short answer. Give it three things: the ability (data and context to push back), explicit permission (instructions that its job is to stress-test), and a cadence (thresholds that trigger a dissenting brief). Without all three you build a yes-machine. Chapters 14, 15 and 38 describe two mechanisms working together. AI trained on your data tends to agree with your worldview, and AI is also bent toward being agreeable before you feed it anything. The book mentions Marc Andreessen building instructions into his prompts telling AI not to agree with him.

Chapter 15 gives the three-part framework. Ability: access to external market data, competitive intelligence, past wrong decisions and diverse internal voices. Permission: tell it in its foundational instructions to stress-test and surface assumptions. Cadence: build challenge into the workflow, such as a dissenting brief on any decision above a capital threshold, anything irreversible, or any people decision involving someone with real institutional history.

The book also warns against the opposite failure: an AI that challenges everything becomes noise. The goal is an AI that knows when to challenge. What to do next
  1. Write the challenge instruction into the AI's foundation document.
  2. Pick two or three decision thresholds that always trigger a dissenting brief.
  3. Review quarterly whether it challenged too often or too rarely.
 Offering: Design & Strategy
Relevant to: CEO, Board and investors

Sources
What is a CEO Doctrinal Statement and how do we write one?
Short answer. A private, specific document describing how you actually decide: about customers, people, risk, growth and your limiters. It is the book's most important pre-implementation task. It takes a weekend and does not require a developer. Chapter 45 distinguishes it from a mission or purpose statement. Those are public-facing and aspirational. A doctrinal statement is private and opinionated: it says how you decide, not what you are trying to do. You cannot train a machine on intuition, only on what is written down.

The book suggests starting with the questions your company faces most. How do you think about customers, and where is the line between accommodation and eroding value? How do you think about people, performance and firing? What is your risk profile? What does healthy growth look like, and what growth would you say no to? Which limiters are real and which are inherited assumptions?

Write five pages or twenty, then give it to the team for thirty minutes of reaction, since pushback forces you to articulate why you believe what you believe. The same process then repeats for the CFO, CRO, COO and every other seat, one at a time. What to do next
  1. Write answers to the five question groups in your own words.
  2. Circulate it to your team for reaction and revise.
  3. Give it to the AI as the foundation document, then repeat for each leader.
 Offering: Design & Strategy
Relevant to: CEO, CFO and finance, COO and operations, Sales leader

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 45; Ch. 42
What is business bias and why would we teach it to AI?
Short answer. In the book, bias means your company's worldview and preferences: how you treat underperformers, how much cash you hold, your risk tolerance and methodology. It does not mean prejudice. A neutral AI is a generic AI, and bias is what makes yours useful. Chapter 39 is careful to say bias here is not discrimination or unfair treatment of people. It is closer to a preference, the way someone reaches for a brand of shoe without thinking. A business runs on preferences like these.

The book's example in Chapter 41 asks how your company handles underperformers: retrain, reposition, fire with severance, fire without. The author states his own answer, and the point is that the impact of the bias shapes how your business decides. It also asks what methodology you run: Scaling Up, Jim Collins, EOS, Great Game of Business or Lencioni. Your AI should know.

Chapter 40 adds that every AI has a bias from training and from choices its builders made, so a neutral AI does not exist. The question is whether the configuration works for you. What to do next
  1. Decide the five to ten preferences that define how you operate.
  2. Record them in plain language.
  3. Feed them to the AI and weight them above generic best practice.
 Offering: Design & Strategy
Relevant to: CEO, HR and people, Board and investors

Sources
What should we tell AI about our company so its answers fit us?
Short answer. Seven things: core values as actually practiced, risk profile, strategic frameworks, tone and voice, red lines, signature moves and customer persona. The book says this takes two to five pages and changes every answer. Chapter 39 lists the seven categories. Core values are the ones that show up when a hard decision has to be made, not the lobby poster. Risk profile tells AI when to push and when to play safe. Strategic frameworks, such as Jim Collins, Scaling Up or Rockefeller Habits, mean AI reasons inside your frame. Tone and voice cover how your company talks. Red lines are decisions you refuse to make. Signature moves are things you are already known for doing well. Customer persona is who your customer actually is, in their words.

Chapter 41 adds more inputs: company history including where it succeeded and failed, culture, purpose, brand promise, one-page plan, BHAG, flywheel, who on the leadership team is strong and who has challenges, industry trends and customer preferences. Past strategies go in for historical context and should be weighted so AI does not treat them as current. What to do next
  1. Draft one paragraph per category.
  2. Add your history, one-page plan and leadership profile.
  3. Mark old strategies as historical so they are weighted down.
 Offering: Design & Strategy
Relevant to: CEO, COO and operations, Owner-operator

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 39; Ch. 41
What is data weighting and why does it matter?
Short answer. Weighting tells AI whose input and which data should count more, instead of averaging every opinion equally. Left alone, AI weights every voice close to the same, which the book calls dangerous. Chapters 36, 43 and 46 build the argument. Humans instinctively know which voices to trust. AI does not, so you make that explicit. Without weighting AI does not reason, it averages, and averaging every opinion is not synthesis. Unweighted data produces noise with confidence.

The book's examples include a planning meeting where the VP of Operations, a new regional manager and a sales VP known for inaccurate forecasts disagree about cutting an expensive extra shift. Nobody had read the contract. AI left to itself hears all three at roughly equal weight. It also uses Air Canada's chatbot, ended up before a tribunal, with an $812 outcome, after giving wrong bereavement-fare information, as a weighting failure: nobody taught it that actual policy outranks a plausible guess.

Wyant calls data weighting a professional skill, like reading a balance sheet, and says almost nobody is developing it. What to do next
  1. Decide which roles carry more authority on which topics.
  2. Tag data by source and speaker.
  3. Test the system with a known disagreement and check whose view it favors.
 Offering: Design & Strategy
Relevant to: CEO, CFO and finance, COO and operations

Sources
How do we decide whose opinions AI should weight more?
Short answer. Base it on track record and relevance to the topic, not title alone. The CFO's read on cost can outweigh a first-year analyst's, without silencing the analyst. Use numbers, not adjectives, to express weight. The book gives examples: the CFO on cost, the CMO on brand, the CEO's intent across all of it, a veteran plant manager over a new hire on supply chain. Chapter 46 says to build weighting deliberately, based on track record and relevance.

It also warns that your loudest person is not always your best person, and that your most confident wrong answer and your quietest right one can land with equal weight if you do nothing.

Chapter 33 explains the equity side. Transcripts capture every word at equal fidelity and do not reward volume, so quiet contributors reach the AI. That does not mean every voice is equal on every decision. Judgment and expertise still matter. What to do next
  1. List the decision areas where weight should differ.
  2. Assign each contributor a weight per area on a fixed scale.
  3. Review weights each quarter as track records change.
 Offering: Design & Strategy
Relevant to: CEO, CFO and finance, HR and people

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 33; Ch. 36; Ch. 46
What is the Golden Window for AI?
Short answer. The period when AI is powerful enough to matter but still open to customization, so any company can build its own intelligence layer. The book says it pays in six ways: data, people, intelligence, bugs, time and cost. Chapter 4 describes the Golden Window as a real, measurable period. Cleaning your data now is cheaper and calmer than later. Starting now gives you a trained staff in a year. The intelligence layer compounds as you feed it. You work out bugs quietly on low-stakes work, since the book compares building the end state to an ERP implementation more than installing an app. What you build now echoes forward. And it is the least expensive it will ever be.

Wyant adds a caveat: moving first carries risk, and some of what you try will not work. You are buying a compounding advantage with a tolerance for a few dead ends. He also says the window will not stay open indefinitely. What to do next
  1. Pick the single data set or process you will start with this quarter.
  2. Set a small budget and accept some dead ends.
  3. Model what one year of delay costs you.
 Offering: Design & Strategy
Relevant to: CEO, Board and investors

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 4; Ch. 55
What does waiting for AI to settle down actually cost us?
Short answer. More than in past technology waves. The book argues AI is one of the only technologies where the tool improves itself, so a late start is not a fixed head start you can close. It is a compounding gap. Chapter 4 contrasts it with enterprise software, cloud and mobile, where a late mover who worked hard could close a fixed head start. With AI, a company that builds a private AI today accumulates data, refinement and institutional intelligence that grow with each quarter. The late mover also chases a target that is moving away.

The author's financial-compounding analogy compares two investors, one starting at thirty and one at forty. He suggests modeling the monthly cost of delay, including competitors' compounding over three years. He admits there is no hard study proving the size, and says some of the return must be taken on reasoned faith. What to do next
  1. Estimate what a competitor's AI advantage could be worth in three years.
  2. Compare it to the cost of a pilot now.
  3. Start the data and enablement work that compounds regardless of vendor.
 Offering: Design & Strategy
Relevant to: CEO, CFO and finance, Board and investors

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 4; Ch. 26
Private AI vs. ChatGPT Enterprise: what's the difference?
Short answer. Enterprise accounts give people a secure interface to a vendor's model. Private AI, as the book defines it, is your own interface, data, workflow logic and institutional knowledge in an environment you control, connected to whichever models fit. It is the same underlying power, configured to know who you are. Chapter 6 explains that ChatGPT is an interface into OpenAI's model, and that vendors expose APIs so you can build your own interface connected to your own data. That is private AI. Chapter 3 warns that a few people with enterprise accounts patting themselves on the back have more confidence than capability, with a weak foundation underneath.

Three differences stand out in the book. Memory and context: enterprise tools remember less than you think and on the vendor's terms, while private AI remembers what you need on yours. Lock-in: Chapter 7 says building a layer that is not tied to one provider costs about ten percent more effort for near-complete independence. Cost: scanning a shared folder with a generic tool costs more as you lean on it, and the cost is not linear.

Wyant says this is not either-or. Use ChatGPT, Claude and Gemini. Just do not weld your company to one. What to do next
  1. List what your enterprise tool does not know about your company.
  2. Ask what happens to your workflows if the vendor changes price or terms.
  3. Start a pilot on a private layer connected to more than one model.
 Offering: Design & Strategy
Relevant to: CEO, IT and technology, CFO and finance

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 3; Ch. 6; Ch. 7; Ch. 8
What is private AI?
Short answer. An AI system built on your own company's data and kept private to your organization, rather than a shared public model. The book says it is not a different AI. It is the same underlying models configured to know who you are. Chapter 6 uses the image of a brilliant friend with amnesia. General AI knows the world but starts over on anything specific to your company. Private AI fixes that with your strategy, culture and decision history. It does not require inventing a new model. The author stresses you do not need to rebuild the large language model, only the interface and data layer on top of it.

Chapter 8 separates three kinds of AI: general-purpose (ChatGPT), guided or vertical (Harvey AI for law, for example), and captured or private. The book recommends using all three, with private AI as the most important. It also stresses that retrieval, a mechanism for searching your own documents, is one building block and not the whole solution. Agents, connectors and other pieces layer on top.

Chapter 6 also says private AI is not significantly harder than what you already use, and that the difference in outcome is enormous while the difference in effort is not. What to do next
  1. Write down the questions a generic AI cannot answer about your business.
  2. Identify the data that would answer them.
  3. Pilot retrieval over that data in a controlled environment.
 Offering: Design & Strategy
Relevant to: CEO, IT and technology, Owner-operator

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 6; Ch. 8
How do we avoid AI vendor lock-in?
Short answer. Keep your interface, data, workflow logic and institutional knowledge in an environment you control, and connect it to whichever model fits. The book says that costs about ten percent more effort and buys near-complete independence. Chapter 7 uses a house-building analogy. Buying from one supplier is simple. Using several costs about ten percent more coordination, but if one supplier raises prices, runs out or fails, the job keeps moving. A private layer lets you swap models without redesigning every workflow.

The risks are concrete. The book notes the November 18, 2025 Cloudflare outage briefly took ChatGPT, Claude and other services offline, with full stabilization taking around six hours. It also lists provider price changes, terms changes and retired models as events that arrive on the vendor's schedule. Its warning: the most expensive sentence in AI strategy may be 'we'll make it portable later.'

Independence does not mean refusing the big vendors. It means that when the market changes, the model can change without rebuilding the company around it. What to do next
  1. Inventory which workflows are built inside a single vendor's platform.
  2. Decide what must live in a layer you own: data, prompts, workflow logic.
  3. Design fallbacks so one provider outage does not stop the business.
 Offering: Design & Strategy
Relevant to: CEO, IT and technology, CFO and finance

Sources
How do we capture what our key people know before they retire or leave?
Short answer. Run a capture sprint on them before they walk out: interviews and transcripts that preserve their reasoning, then structured into your AI. The book estimates you would otherwise retain about one percent of what made their judgment valuable. Chapter 6 calls AI a succession plan, a living will of leadership. Most intelligence that runs a company lives in a few heads and in spoken conversation, not documents. Without capture, if your three best leaders left tomorrow, the book's blunt estimate is you would keep about one percent of what they know.

Chapter 44 gives the image of a top salesperson retiring: cake, a card, applause, and every client relationship and avoided mistake walks out. The book says to run the capture sprint before she leaves, and the workbook (Section 3.7) lays out who to interview and in what order.

It is realistic about limits. The result is not a perfect clone, but it is enough to cut a successor's learning curve substantially. What to do next
  1. List the five people whose departure would hurt most.
  2. Schedule recorded interviews on decisions, not just processes.
  3. Structure the transcripts and load them into your AI with weighting.
 Offering: Design & Strategy
Relevant to: CEO, HR and people, COO and operations, Manufacturing, PE-backed and family-owned, Trades and field services

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 6; Ch. 44
We're in a regulated industry. Is compliance a reason to wait on AI?
Short answer. The book says no. Compliance is a parameter for how you implement AI, not a stop sign. Your competitors face the same rules and are moving carefully, and most of your company is not regulated data anyway. Chapter 27's line is that compliance is a lane marker, not a stop sign. Its advice has three parts: start early, experiment carefully, and challenge the people who say it cannot be done. You can learn the tools using synthetic data, de-identified data, last year's closed quarter or a scrubbed test set before touching regulated records.

It also points out that in most regulated companies the actual regulated data is a small fraction of weekly work. Scheduling, internal meeting notes, vendor emails, HR questions, onboarding and policy drafting are not PHI and not SOX controls, so they can move into private AI first.

The author's suggestion is to bring legal and compliance in early as architects who define parameters, such as what data AI can access, what is off limits and what approval workflows must exist, rather than as gatekeepers. What to do next
  1. Split your use cases into regulated and unregulated data.
  2. Pilot on the unregulated side or on synthetic data.
  3. Have compliance help define access rules before the build.
 Offering: Design & Strategy
Relevant to: CEO, Legal and compliance, CFO and finance, Financial services, Healthcare

Sources
What do HIPAA and SOX actually require when we use AI?
Short answer. Neither bans AI. HIPAA treats an AI tool touching protected health information like any vendor that touches it: a business associate agreement, encryption, access control and audit logging. SOX asks whether AI touches anything that feeds financial reporting and requires control discipline there. Chapter 27 says HIPAA has no AI exemption. Any AI tool touching PHI is a business associate, needs a signed agreement before processing a single record and must meet the same encryption, access and audit-logging standards. Consumer-grade assistants, such as a free version downloaded on a personal phone, are not HIPAA-compliant by default. An enterprise, contracted version might be, and you find out by reading the contract.

SOX works on the same logic aimed at a different target. A tool that summarizes internal meetings is a very different risk from one that automates journal entries or flags exceptions in controls. Where AI touches the reporting pipeline, internal audit must be able to explain what the model does, test it for drift and error like a human-run control, and keep a human in the loop where SOX required a sign-off.

The book frames both as configuration questions: where does data live, who can touch it, and can a human still explain and approve what happened. It points to on-premises builds or a properly contracted Azure environment as ways to keep data inside an agreement you control.

This is the author's practitioner reading, not legal advice. Confirm with counsel. What to do next
  1. Map each planned AI use to PHI or financial reporting exposure.
  2. Check contracts and business associate agreements for every vendor.
  3. Define human sign-off points for AI-assisted controls.
 Offering: Design & Strategy
Relevant to: Legal and compliance, CFO and finance, IT and technology, Financial services, Healthcare

Sources
What do the EU AI Act and US state AI laws mean for a mid-sized company?
Short answer. The book's central point is explainability. Article 86 of the EU AI Act gives people affected by AI-assisted decisions the right to a clear explanation, and US states are building their own rules. Companies that document AI's role now will be ready. Chapter 18 explains Article 86: anyone affected by an AI-assisted decision in areas such as hiring, credit, healthcare and education can seek a meaningful explanation, and the obligation falls on the deploying company, not the vendor. The book says it does not reach business strategy, only decisions with direct human consequences. After two years of preparation, industry asked in June 2026 for sixteen more months. The compliance date cited is December 2, 2027.

In the US, the book says all fifty states introduced AI-related legislation by 2025 and 145 bills were enacted that year. Colorado passed the first comprehensive governance law and has revised it since. New York City requires annual bias audits for automated hiring tools, and Illinois requires notice and consent before AI analyzes a video interview. There is no comprehensive federal law, which the author calls a patchwork that changes quarterly.

His advice: build governance, documentation and an audit trail now. Regulation, he says, gives existing undocumented AI use a legal address. This is not legal advice. Confirm current rules with counsel, since they change. What to do next
  1. Identify AI use in hiring, credit, insurance, healthcare or similar decisions.
  2. Document the prompt, output and human decision for those uses.
  3. Assign someone to track legal changes each quarter.
 Offering: Design & Strategy
Relevant to: Legal and compliance, HR and people, CEO, Financial services, Healthcare

Sources
If we were served a subpoena for our employees' AI conversations, could we produce them?
Short answer. For most companies the answer is no. Free personal accounts are out of reach, and even enterprise accounts raise unsettled questions about the AI's role. The book recommends building auditable AI and an audit trail before anyone asks. Chapter 18 opens with this question. Most leaders open a tab, ask, use the answer and close it, and the trail evaporates. Email has thirty years of retention policy, legal holds and discovery tools. AI has no framework or norm yet. If conversations happened on personal free accounts, you have no access. If you have enterprise accounts that save history, you still face unsettled questions about where AI's role ends and the human's begins, and at what point 'mainly based on AI output' triggers accountability.

The book describes two plausible moments: someone photographs a screen showing AI involved in a decision, or a manager mentions AI while terminating an employee and it is recorded. A company without a documentation trail is in a serious position.

This is the author's view on a developing area. Ask counsel what your retention duties are. What to do next
  1. Find out where employees' AI conversations currently live.
  2. Move sanctioned work into accounts or a private layer that logs and retains.
  3. Define when AI involvement in a decision must be recorded.
 Offering: Design & Strategy
Relevant to: Legal and compliance, CEO, IT and technology

Sources
Why should we record and transcribe leadership meetings?
Short answer. Because most of your best thinking is spoken, and AI cannot learn from what it never captured. The author's A/B tests found transcripts far superior to notes, and the book calls meetings the most underused data asset in most companies. Chapter 31 describes how the team discovered it. Early data gathering used strategy documents, notes, emails and financials, and it was stale. In a time crunch they fed in transcripts of recorded interviews, results felt fresh and sharp, and A/B testing confirmed it. Wyant's theory is that word choice, order and emphasis change meaning, and notes summarize and lose some of it.

He argues that without recording, AI never sees the human debate and idea generation that make meetings valuable, and that in ten to twenty years you will see AI in every meeting while humans stay in the room. The main barrier, he says, is the stigma around recorded meetings.

Chapter 33 adds that transcripts do not reward the loudest voice, so quieter contributors reach the AI. Chapter 6 says to start the transcript habit: record leadership calls and feed the AI. The book also says this requires planning: tag speakers, weight inputs and get consent as your policies and local law require. What to do next
  1. Pick one recurring leadership meeting and record it for a quarter.
  2. Tag speakers so weighting can be applied.
  3. Set a retention and access policy before you start.
 Offering: Design & Strategy
Relevant to: CEO, COO and operations, Owner-operator

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 6; Ch. 31; Ch. 33
What is shadow AI and how do we get it under control?
Short answer. Shadow AI is unsanctioned AI adoption, including working software built by people with no engineering background that nobody in governance knows exists. The book says to ask where it lives, whether it is secure, who maintains it and who can troubleshoot it. Chapter 3 says AI rolled out in a disjointed way almost everywhere: some handed to IT, some left wide open, with no policies, procedures or design behind it. The book says companies have moved past shadow IT into shadow AI running the company with no controls and unmapped dependencies. Chapter 1 adds that leaders often adopt it on their own and their teams run different versions on different data.

Chapter 30 covers vibe coding: describing what you want in plain language and having AI write the software. It is a real productivity unlock, but the book calls much of the output a one-way ticket. It asks four questions to put to anything built this way. Where does it live? A chat interface or temporary environment is not infrastructure. Is it secure? AI writes functional code quickly but not automatically hardened code. Who maintains it when the technology changes? Who has the context to troubleshoot once the builder has moved on?

The book's stance is not to stop building. Ask the questions before you build, not after. What to do next
  1. Survey teams for AI tools and homemade software in use.
  2. Set an intake rule: anything built with AI gets an owner and a home.
  3. Review security on anything touching company data.
 Offering: Design & Strategy
Relevant to: CEO, IT and technology, Legal and compliance

Sources
Why do leaders automate the wrong things first?
Short answer. The book's observation is that people automate pain, tolerate ease, fight difficulty and protect identity. Leaders rush to automate work they hate and resist letting AI help with hard work they enjoy and are proud of, such as evaluating people or diagnosing problems. Chapters 22 and 23 describe four quadrants of automation behavior. Easy work that nobody minds gets automated slowly. Easy work everyone hates gets automated fast. Hard work people feel neutral about stays by hand. Hard work people enjoy is protected hardest because it has become part of how they see themselves. The author says this is his observation from years of watching leaders, not a peer-reviewed finding.

His examples are Air Canada, which automated a hated call-center task and got sued over a wrong bereavement-fare answer, and doctors who keep diagnosing by hand while billing and check-in get automated without a fight. He suggests executives apply it to themselves: research reviewed in Nine Lies About Work finds people rate others less objectively than they believe, yet executives resist using AI to help assess staff because they enjoy it.

His advice is to use AI at the C-level on tasks that are hard for humans and easy for AI, rather than only on the easy tasks you dislike. What to do next
  1. List the five tasks your leaders dislike most and the five they are proudest of.
  2. Ask which of each AI could help with.
  3. Pilot AI on one pride task with human review.
 Offering: Design & Strategy
Relevant to: CEO, COO and operations

Sources
Where does AI ROI actually show up in the P&L?
Short answer. The book says the serious money is in automation connected to a decision layer, not the chat window. Its worked example is a ten-hour-a-week reporting process at $60 an hour, about $30,000 a year per process, multiplied across many processes. Chapters 48 to 50 argue that most implementations stop at the intelligence layer, the chat window and Q&A. That delivers value but leaves most of the return on the table. Automation without a decision layer gives you speed with no direction: faster versions of the wrong call.

Chapter 49 gives the math. A process requiring ten hours a week to gather data from three systems and produce a leadership report, performed by someone at $60 per hour fully loaded, costs $600 a week, or $30,000 a year. Automated, that time is recaptured or reallocated. Multiply across five or ten processes and it becomes structural change.

Chapter 50 is more cautious. It says the dollar figures are real but that setup is harder than it looks and many companies are still using AI the way they did eighteen months ago. The earlier chapters also admit the largest returns, such as compounding intelligence, are long-term and hard to prove today. What to do next
  1. Inventory recurring reports and status-gathering work in the leadership team.
  2. Cost each at fully loaded hourly rates.
  3. Automate the top candidate and measure the hours recaptured.
 Offering: Design & Strategy
Relevant to: CFO and finance, CEO, COO and operations

Sources
How do we build a data culture?
Short answer. Four moves from the book: decide what matters, make capture non-negotiable, police quality, and play the long game. Collecting data is passive. A data culture is a set of leadership behaviors that treat information as a living asset. Chapter 37 compares unmanaged data to a family garage nobody cleans. You need both the rule and someone enforcing it. A data culture starts with leadership: if the CEO does not treat data as strategic nobody will, and if teams argue over whose numbers are right, you have a data culture problem.

The four moves: first, decide what matters by asking which decisions you make most often that have the highest impact, since that tells you what to capture. Second, make capture non-negotiable, because hallway decisions and unrecorded calls are knowledge walking out the door. Third, police the quality with owners and standards, because bad data is worse than no data. Fourth, play the long game: clients who started collecting transcripts saw extreme value in as little as six months, and it keeps compounding.

Chapter 36 adds data stewards in every department, framed as leadership roles, not IT roles. What to do next
  1. Name the ten decisions that matter most and the data behind each.
  2. Assign a data steward per department.
  3. Set capture rules for meetings, decisions and customer feedback.
 Offering: Design & Strategy
Relevant to: CEO, COO and operations, IT and technology

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 34; Ch. 36; Ch. 37
Why is our data, not the AI model, our competitive advantage?
Short answer. Competitors can buy the same AI tools. They cannot buy fifteen years of your customer relationships, pricing decisions and operating knowledge. The book calls the data the moat and says the advantage compounds each quarter. Chapter 34 says the intelligence layer is commoditizing, OpenAI, Anthropic and Google sell to everyone, so a strategy that depends on a better model has a short shelf life. What cannot be licensed is your history: customer patterns, pricing outcomes, transcripts, methodology and market understanding.

The compounding point: every quarter adds proprietary data and the AI gets better at finding signal in what it already holds. The book cites MIRATECH, where years of leadership transcripts and notes built an advantage month over month, and Rae Corp, where after nine months and about a thousand transcripts the compounding value could be tested.

Wyant's phrasing is that you are not an AI company, you are a data company, and that most time with AI should go to fixing data, not prompts. What to do next
  1. Inventory the proprietary data your competitors cannot get.
  2. Prioritize capturing it consistently.
  3. Treat data quality as a leadership topic, not a back-office one.
 Offering: Design & Strategy
Relevant to: CEO, CFO and finance, Board and investors

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 34
How can an owner-led company grow past 25 employees with AI?
Short answer. The book says AI will not remove the politics that plateau companies near 25 people, but it gives owners tools to manage past it: shared context, process and aligned leadership. Scaling is still a choice. Chapter 11 describes a common plateau around 25 people, where drama, competition and too many opinions slow growth. It cites Les McKeown's Predictable Success and the whitewater stage. Some owners stop there on purpose, running what YPO calls a lifestyle business.

Wyant's argument is that AI makes the choice to scale survivable: it gives owners real tools for challenges that used to make scaling past that size too hard. He expects AI to push the plateau higher, perhaps to 100 or even 500 employees, including for second and third generation family businesses. He also sees opportunity in construction or services firms stuck at 10 to 15 people as acquisition targets for operators who can fold them into a larger family of companies using process and AI.

He is clear it does not remove drama and can make it worse if everyone gets a confident AI argument for their own side, which is the case for a shared brain. What to do next
  1. Decide whether you want to scale past the plateau.
  2. Put shared context and process in place before adding headcount.
  3. Use one shared AI environment so leaders argue from the same facts.
 Offering: Design & Strategy
Relevant to: Owner-operator, CEO, PE-backed and family-owned, Trades and field services

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 11; Ch. 14
  • Predictable Success, Les McKeown
What is IT's role in an AI project?
Short answer. IT owns the plumbing and the guardrails: enterprise accounts, access, security, integrations and securing automation that can touch systems. It should not own strategy. The book says this is the one place it wants IT in the room. The book tells leaders throughout to keep IT out of strategy, then says in Chapter 50 that when automation can touch systems, move money or change records, securing it is IT's job and they should be brought in. Chapter 51 says verifying that base tools are set up right, with real enterprise accounts, adoption, collaboration and connections to data, is the first place IT drops the ball and probably the single thing IT should be doing.

Chapter 25 adds that the people best equipped to lead implementation are business analysts and strategists, not programmers, because the technical side is easier than teaching the system how leadership thinks. Chapter 50 cites Gartner's projection that by 2028, 45 percent of CIOs will lead AI agent systems operating outside IT. What to do next
  1. Give IT ownership of identity, access, logging and integrations.
  2. Have IT review every automation that can change records.
  3. Keep use-case selection with business owners.
 Offering: Design & Strategy
Relevant to: IT and technology

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 25; Ch. 50; Ch. 51
  • AI Agent Layer, Gartner
What should general counsel ask about AI use in the company?
Short answer. Where AI conversations are stored, whether the company could explain an AI-assisted decision affecting a person, which tools touch regulated data, and who owns documentation. Counsel should act as architect of the parameters, not gatekeeper. Chapter 18 poses the central questions: could you produce AI conversations in a subpoena, can you show the original prompt and response in sequence, and where does AI's role end and a human's begin. It points to Article 86 of the EU AI Act, the patchwork of US state laws and the governance decisions leadership must make.

Chapter 27 adds the regulated-industry view: HIPAA business associate agreements, encryption, access controls and audit logging, and SOX control discipline where AI touches financial reporting. It recommends involving legal and compliance early to define what data AI can access, what is off limits and what approval workflows must exist.

These are the author's practical framing, not legal advice. Counsel should apply current law. What to do next
  1. Ask for an inventory of AI tools and where data goes.
  2. Define retention and documentation requirements for AI-assisted decisions.
  3. Review vendor contracts for data use and training terms.
 Offering: Design & Strategy
Relevant to: Legal and compliance, Financial services, Healthcare

Sources
What should HR know before using AI in hiring and performance reviews?
Short answer. AI-assisted decisions about hiring, firing and similar matters are exactly where the book sees the most legal exposure. Document AI's role, know your jurisdiction's rules, and keep a human accountable. The book also argues AI can help people evaluate staff more consistently. Chapter 18 explains that the EU's Article 86 targets decisions with direct human consequences, including hiring and firing. In the US, New York City requires annual independent bias audits for automated hiring tools and Illinois requires notice and consent before AI analyzes a video interview. If a manager says out loud that AI was part of a termination and it is recorded, the company needs a documentation trail.

Chapter 23 argues that humans rate others less objectively than they believe, and that executives resist AI-assisted evaluation partly because they enjoy evaluating. Chapter 36 describes the weighting problem: who the AI should believe about an employee, the manager, peers or customers.

Chapter 49 describes AI taking over manual sifting of candidate packets for the head of HR. This is the author's framing, not legal advice. What to do next
  1. List every AI use in hiring, reviews and terminations.
  2. Record the AI's input and the human decision for each.
  3. Check local rules for audits and notices before use.
 Offering: Design & Strategy
Relevant to: HR and people, Legal and compliance

Sources
How can a healthcare organization use AI without violating HIPAA?
Short answer. Treat any AI tool that touches protected health information as a business associate with a signed agreement, encryption, access control and audit logging, and move non-PHI work first. The book says most of a healthcare company's week is not regulated data. Chapter 27 sets out the HIPAA requirements and warns that consumer assistants downloaded on a personal phone are not compliant by default. It also stresses that scheduling, internal meeting notes, vendor emails, HR questions, onboarding and policy drafting are not PHI and can move into private AI first. You can learn the tools on synthetic or de-identified data.

The appendix gives a healthcare group example where thousands of uploaded documents produced contradictions until terminology was standardized, documents tagged by department and outdated files removed. Chapter 13's IBM Watson Health story and Chapter 22's discussion of doctors keeping diagnosis in human hands show the book's view: AI helps, but trust and judgment remain human.

This applies the book's general guidance to healthcare. It is not legal advice. What to do next
  1. Separate PHI use cases from the rest and start with the rest.
  2. Confirm business associate agreements for every tool.
  3. Clean and tag clinical and policy documents before ingestion.
 Offering: Design & Strategy
Relevant to: Legal and compliance, CEO, IT and technology, Healthcare

Sources
How should a financial services firm use AI without handing it decisions?
Short answer. Design AI as a decision supporter, with guardrails, human sign-off where regulation requires it, and testing. The appendix describes a financial services firm whose AI sounded too confident until guardrails shifted it from decision maker to supporter. The appendix example is direct: the AI retrieved the right data but made confident recommendations in areas needing human judgment and regulatory review. Guardrails fixed it, and the book calls that the right design. Chapter 27 adds SOX discipline: explain the model, test for drift and error, and keep human sign-off wherever it has always been required.

Chapter 10's Zillow example shows the cost of acting on confident numbers no human checks, and Chapter 11's banker CEO shows the opposite failure of checking three models endlessly. Chapter 18 notes credit and insurance decisions fall within the EU Article 86 explainability right.

This applies the book's general guidance to financial services. It is not legal or compliance advice. What to do next
  1. Mark which outputs are recommendations and which are decisions.
  2. Require human sign-off and logging for the latter.
  3. Test regularly for drift and overconfidence.
 Offering: Design & Strategy
Relevant to: CFO and finance, Legal and compliance, CEO, Financial services

Sources
How does AI change staffing in professional services firms?
Short answer. The book reports junior analytical and administrative roles shrinking because a senior person with good data and AI can do more without a bench underneath. The differentiator is clean data, not just intelligence. Chapter 36 observes that paralegals and junior analysts at firms like KPMG are starting to disappear, not because the work vanished but because a senior person with domain knowledge, paired with AI, can do far more. The author says the same is true at Red Wagon Advisors, a sister company of REDEGADES.AI: fewer junior people are needed because senior people deliver a better product alone.

He adds that it is not about how smart the senior person is. A brilliant consultant without data on a specific company cannot do much. The same person with good clean data no longer needs a room of juniors building the proof and the presentation. Smart plus clean data changes the game.

Chapter 8 notes vertical tools like Harvey AI for law, and Chapter 10's lawyer with six invented cases shows the checking duty. Chapter 8 also says to think about careers: some industries will need fewer people. What to do next
  1. Map which junior tasks are research and assembly.
  2. Invest in client and firm data quality first.
  3. Redesign career paths around judgment and client work.
 Offering: Design & Strategy
Relevant to: CEO, HR and people, Owner-operator, Professional services

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 8; Ch. 10; Ch. 36
Is AI relevant to a trades, construction or other 'unglamorous' business?
Short answer. The book says yes, and that the quiet revolution is happening in businesses like duck farms, coolant makers, recyclers and fire-protection installers. AI augments them, and no software replaces a duck farm. Chapter 55 describes real clients: a duck farmer in Indiana, a coolant company in Tulsa, a hydroponic lettuce farmer in Massachusetts, recyclers in Wisconsin and a company installing fire detection systems. They are businesses built to last a hundred years, and AI augments them. Chapter 9 also lists agriculture, tech recycling and clean energy among industries served.

Chapter 11 sees opportunity in construction and services businesses stuck at 10 to 15 people as acquisition targets that can be folded into larger groups using process and AI. Chapter 55's advice is to start capturing transcripts, begin with the leadership team, and build one shared brain. What to do next
  1. Pick the single decision your owner makes most often.
  2. Record how the owner makes it for a month.
  3. Use that record as your first AI context.
 Offering: Design & Strategy
Relevant to: Owner-operator, CEO, Trades and field services

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 9; Ch. 11; Ch. 55
How can PE-backed and family-owned companies use AI for succession and roll-ups?
Short answer. Use it to capture leadership knowledge before transitions and to bring acquired companies onto shared process faster. The book calls private AI a succession plan and sees AI pushing the size plateau for family businesses higher. Chapter 6 argues the most valuable asset in the company is the thinking that built it, and that private AI can encode a leader's reasoning so a successor's learning curve drops. It extends this to every key person, not only the CEO. Chapter 3 adds that a new executive can query the system and absorb years of context in weeks.

Chapter 11 expects AI to push the plateau higher, including for second and third generation family businesses, and sees roll-up opportunities for operators who can combine process with AI. Chapter 9 lists private equity-backed growth companies among the author's clients. Chapter 26 offers AI as an extra partner alongside human partners, who inevitably change course. What to do next
  1. Identify the leaders whose knowledge is least documented.
  2. Run capture interviews and load them into a shared environment.
  3. Use the same capture method when onboarding an acquisition.
 Offering: Design & Strategy
Relevant to: CEO, Board and investors, Owner-operator, PE-backed and family-owned

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 3; Ch. 6; Ch. 9; Ch. 11; Ch. 26
How do we use AI on our own documents without exposing data?
Short answer. Build the AI inside an environment you control, such as a properly contracted Azure tenant, with retrieval over your own files and access rules enforced in the system. The book says this keeps data inside your walls and out of third-party training data. Chapter 27 says private AI deployed on enterprise-grade infrastructure like Microsoft Azure keeps your data inside your environment: it does not leave your walls and is not sent into a third-party AI company's training data. In many cases it reduces compliance risk because the system enforces data access rules programmatically instead of relying on human discipline.

The mechanism is retrieval-augmented generation. Your documents are cleaned, tagged, broken into chunks and stored so the AI searches them when it answers, and it can cite where an answer came from. Chapter 8 notes you can point the AI at one document, a set of documents, or a mix with public information, which gives you control over exactly what it examines.

The appendix adds the operational controls: set user roles and document permissions, decide what the C-suite sees first and what stays off limits, keep logging and an audit trail, and confirm permissions and security when connecting SharePoint, OneDrive or Google Drive. For regulated data, read contracts rather than assuming. Consumer AI apps are not HIPAA-compliant by default. What to do next
  1. Classify documents: regulated, confidential, general.
  2. Choose a hosting environment you control and review the vendor contract on data use.
  3. Set roles and document permissions before loading anything.
  4. Start with non-regulated documents while you test.
 Offering: Implementation
Relevant to: CEO, IT and technology, Legal and compliance, CFO and finance, Financial services, Healthcare

Sources
What is RAG, in plain English?
Short answer. Retrieval-augmented generation lets your AI search your own company data when it answers. It is like a brilliant MBA graduate who has now read every memo, report and plan you have written. The book calls it one building block, not the whole solution. The appendix explains RAG as how you give your AI access to your information instead of everyone else's. A standard AI is like an MBA who has read everything published but never set foot in your company. A RAG-powered AI has also read your memos, sales reports and strategic plans. Asked why margins dropped last quarter, a generic AI talks about supply chains and market conditions. With RAG it pulls your financials, sales data and operating notes from that period.

The book's framing is that you are not building a smarter AI, you are building a system that refuses to answer without checking your sources first. Chapter 8 stresses RAG is one mechanism among several. A private AI can also include agents, data connectors and other layers. It also says you can direct the system to weight a specific document, such as a one-page plan, more heavily. What to do next
  1. Pick one question your team asks repeatedly that a generic AI answers badly.
  2. Gather the documents that contain the real answer.
  3. Test retrieval on that one question before adding more.
 Offering: Implementation
Relevant to: CEO, IT and technology, Owner-operator

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 8; Appendix
How do we build a private AI, step by step?
Short answer. The appendix gives five build steps (gather what matters, clean and structure, create retrieval, layer guardrails, make it think like you) plus a nine-part checklist from picking a use case to optimizing. Start with one problem. The book's five conceptual steps: gather the knowledge that matters, asking what you would lose if your best employee quit tomorrow; clean and structure the data through normalization and tagging; create the retrieval system; add guardrails; and shape how it thinks through weighting and encoded leadership philosophy.

The checklist then runs in nine stages: pick the right use case; pick your support, with a business owner and a technical owner; pick platforms (hosting such as Azure, AWS or Google Cloud, front end, RAG framework such as LangChain or LangFlow, LLMs, storage, embedding model, security model and user access rules); subscribe and configure; connect to document sources; configure settings and build the interface; normalize and import data; test and refine; and optimize.

The short version in the book: pick one business problem, get technical help, pick your stack, subscribe, connect systems, build the interface, clean and load data, test with real questions, refine, and maintain. What to do next
  1. Write the one-sentence use case and the success test.
  2. Name the business owner and the technical owner.
  3. Work the checklist in order and do not skip testing.
 Offering: Implementation
Relevant to: CEO, IT and technology, COO and operations

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Appendix
Where should we host a private AI: Azure, AWS or Google Cloud?
Short answer. The book recommends starting where your team already lives, and says that for most mid-market companies that is Azure, for four reasons: familiarity, security, AI positioning and business alignment. AWS or Google Cloud can be the right call for some teams. The appendix's reasons for Azure: Microsoft is already inside most enterprises through Microsoft 365, Teams, SharePoint and Outlook; Azure is built with enterprise security and compliance at its core and ties into enterprise identity, which supports governance; its partnership with OpenAI gives native access to widely used models; and it fits existing contracts and workflows, with possible leverage from existing licensing.

It is balanced about alternatives. AWS is powerful but a flexible engineering platform first and puts more configuration responsibility on the company, which is fine for deep technical teams already running it. Google Cloud is an excellent option for teams embedded in Google Workspace. The principle is to start where your team already lives.

This is the REDEGADES.AI team's practitioner view, drawn from client builds. What to do next
  1. Inventory your existing cloud and identity systems.
  2. Check licensing you can reuse.
  3. Review security and compliance fit with your industry.
 Offering: Implementation
Relevant to: IT and technology, CEO, Financial services, Healthcare

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Appendix; Ch. 27
How much does a private AI cost to run?
Short answer. The appendix gives typical subscription ranges: a small pilot in the low hundreds to low thousands per month, a mid-size internal RAG in the low thousands, and larger multi-user systems several thousand per month or more. Implementation effort is separate. The appendix lists typical subscription categories: cloud hosting, LLM usage, vector database, workflow and orchestration tools, monitoring and logging, and security. Typical ranges are a small pilot at low hundreds to low thousands of dollars monthly, a mid-size internal RAG at low thousands, and larger multi-user systems at several thousand a month or more.

Chapter 8 adds a cost warning about the cheap alternative. Pointing ChatGPT or Claude at a shared company folder works and costs nothing to set up, but each scan has a cost that is not linear and climbs as you lean on it, so a private build can become cheaper over time. Chapter 7 says building independent of any one provider takes roughly ten percent more effort than plugging into one platform.

The book's own figures cover subscriptions only, so budget separately for data cleanup, design, enablement and testing. What to do next
  1. Estimate the subscription tier for your pilot.
  2. Add internal time for data preparation and testing.
  3. Compare against the cost of per-scan usage on a generic tool.
 Offering: Implementation
Relevant to: CFO and finance, CEO, Owner-operator

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 7; Ch. 8; Appendix
How do we pick our first AI use case?
Short answer. Pick one problem that matters, has clear value if solved, and is currently handled by educated guessing. The appendix says not to start broad, and to prefer work needing subjective or strategic thinking where standard software struggles. The appendix's checklist says to pick one part of the business that needs deeper customization, start with a real constraint or recurring problem, avoid starting too broad, and choose a case with clear value. Its pro-tip is to start with something requiring subjective or strategic thinking for a problem you are unsure how to solve with standard software, because that is where AI wins.

The book's cautionary view is that the most common mistake is starting too big and trying to transform the organization in a single project. Chapter 44 suggests the 90-day shape: by day 60 one real use case is live in front of one real executive, being torn apart weekly, and by day 90 you stand in the boardroom with results.

Chapter 51 also advises starting with the easy things, not the hard ones. What to do next
  1. List three candidate problems and score value and data availability.
  2. Choose one with an executive sponsor.
  3. Define what success looks like in 90 days.
 Offering: Implementation
Relevant to: CEO, COO and operations, Owner-operator

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 44; Ch. 51; Appendix
Who should own an AI build inside the company?
Short answer. Two owners: a business owner who understands the goal and a technical owner who can build toward it. The appendix says business and technical people who do not work together almost always fail, and that you must also decide who maintains it after launch. The appendix says ownership has to be real, not theoretical. Leadership assumes AI will figure itself out, and tech assumes leadership will define what they need. Neither holds. If you are not a developer, get technical support; the pro-tip says a technical partner saves significant time and prevents costly mistakes.

The appendix's list of what humans must do is leadership work: decide the use case, pick platforms, approve access, choose documents, clean data, set security rules, test answers, approve output, keep data fresh and own the system after launch. AI can do much of the implementation labor, such as writing code, pipelines and tests, but the judgment is yours. Chapter 25 adds that strategists and business analysts often make better leads than programmers. What to do next
  1. Name both owners in writing.
  2. Assign a maintainer for after launch.
  3. Set a weekly review between business and technical owners.
 Offering: Implementation
Relevant to: CEO, IT and technology, COO and operations

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 25; Appendix
How do we prepare our data for AI?
Short answer. Curate before you ingest: decide what goes in and stays out, remove duplicates and outdated drafts, tag and organize by type, date and unit, define source-of-truth documents, and then import. The book says relevance matters more than volume. Chapter 35 and the appendix agree: AI amplifies what you give it, so clean first. The appendix checklist includes choosing what stays out, removing duplicates, junk, outdated drafts and conflicting files, tagging important documents, organizing by type, date and business unit, and defining source-of-truth documents.

The appendix example is a healthcare group that uploaded thousands of documents and got contradictions, with policy conflicts and outdated material beside current. After standardizing terminology, tagging by department and removing outdated files, the same AI became trusted. Same AI, different data.

Chapter 36 states the principle of Structure Before Scale, and says AI does not need massive data, it needs the right data, structured well. The line to remember: AI is not a janitor, it amplifies the mess. Chapter 35 also warns that AI bailing you out of bad data makes you less motivated to fix it. What to do next
  1. Run a data inventory by department.
  2. Pick source-of-truth documents and archive the rest.
  3. Standardize terminology and tags before import.
 Offering: Implementation
Relevant to: CEO, IT and technology, COO and operations, Healthcare

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 16; Ch. 35; Ch. 36; Appendix
Why does AI give confident but wrong answers on our own data?
Short answer. Usually the data: outdated, duplicated, conflicting, dense or simply wrong. The book's MIRATECH team saw confident wrong answers until they fed the right things instead of everything. Chapter 16 tells the story. The team ingested heavily, trusted outputs and got confident, wrong answers. A 200-page KPMG report defeated the AI because of its density and format, though a human could correlate it easily. Sometimes the data was flatly wrong. The breakthrough came from stopping feeding it everything and feeding it the right things aligned with structured data like board reports, KPIs and strategic goals.

Chapter 36 adds that the difference between brilliance and failure is usually invisible to the human eye and almost always comes back to data structure. Brilliance is the exception, not the baseline. Chapter 10 pairs this with a rule: put checkpoints in place regardless. What to do next
  1. Test with questions whose answers you know.
  2. Trace wrong answers back to the source documents.
  3. Remove or fix conflicting and outdated files, then retest.
 Offering: Implementation
Relevant to: CEO, IT and technology, CFO and finance

Sources
How do we test an AI system for hallucination and degradation?
Short answer. Keep a standard list of real business questions, ask them every round, and check correctness, usefulness, sourcing and security. Include fake names and numbers that do not exist in your data and confirm the system says it does not know. The appendix's test step: write a standard test list of real business questions, ask the same ones each round, check accuracy, usefulness, sourcing and security, check that the right documents are pulled, check hallucination risk, compare outputs across models if needed, refine prompts, retrieval, tags and access controls, and repeat until answers are reliable. It says testing is where the truth shows up, and that it is the first step cut under deadline pressure.

Chapter 10 describes REDEGADES.AI running a weekly test on each client system, feeding fake names, products and numbers and watching whether it says it does not know. The book reports the systems say they do not know every time. That is REDEGADES.AI's own client testing and not independent evidence, and the book still advises human checkpoints on decisions.

Chapter 35 adds that AI can degrade over time based on what it is fed or how retrieval was built, so set up recurring tests for degradation and consistency. What to do next
  1. Write 25 to 50 real questions with known answers.
  2. Add decoys that should produce 'I don't know'.
  3. Run the set after every change and on a schedule.
 Offering: Implementation
Relevant to: IT and technology, CEO, Legal and compliance

Sources
Can we start with a shared company folder instead of a full private AI?
Short answer. Yes, as a running start. Save a universal company folder where everyone can reach it and point every AI tool at it. The book warns that per-scan costs climb and that this is a small part of what private AI offers. Chapter 8 describes saving a company resource folder on SharePoint or a shared drive and pointing every AI tool people use at it, so everyone works from the same strategy documents, positioning and institutional memory. It costs nothing to set up.

The book is clear about limits. Each time ChatGPT or Claude scans the folder there is a cost, and it is not linear: the more you lean on it, the better the intelligence and the faster the cost climbs. A folder and a Claude.md file are a running start, not the destination, and the book says they represent a small percentage of real private AI capability.

The same chapter says many readers have already pointed a tool at a folder and watched it get smarter, which is real and worth doing. What to do next
  1. Create the folder with a clear structure and owner.
  2. Include your one-page plan, values, positioning and key policies.
  3. Track usage cost and decide when a private build pays off.
 Offering: Implementation
Relevant to: CEO, IT and technology, Owner-operator

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 8
Is a multi-LLM setup worth it for a mid-sized company?
Short answer. The book says yes: use the best model for each job, switch when price or quality changes, and keep fallback logic if one fails. It describes running ChatGPT, Gemini, Claude and Grok from one private layer. Chapter 7 says independence means your interface, data, workflow logic and knowledge live in an environment you control and can connect to whichever model fits. If one becomes better, use it, if another becomes cheaper, switch, if one becomes unreliable, move. Chapter 48 describes a multi-LLM architecture running several models and switching as the task demands.

The appendix checklist includes fallback logic if one model fails and multi-model logic if comparing outputs. The book says this costs about ten percent more effort. It also reports the author's view in Chapter 55 that the ground moves: a year earlier he thought OpenAI was clearly winning and now sees Anthropic as having closed the gap. What to do next
  1. List the tasks where model choice matters.
  2. Build the interface so models are swappable.
  3. Re-evaluate model choice every six months.
 Offering: Implementation
Relevant to: IT and technology, CEO, CFO and finance

Sources
What are AI agents and MCPs, and how do we automate safely?
Short answer. Agents are AI tools that go do the work, not just answer. MCP, the Model Context Protocol, is the wiring that lets AI plug into your email, CRM and project tools. The book says move fast on simple connections and slow on governance and security. Chapter 50 sorts the terminology. MCP is the wiring, the protocol that lets AI plug into tools you already run. Agents are tools that act: the author had one reply to LinkedIn messages and accept invitations in a browser. Chapter 48 compares MCP to an HDMI cable, unglamorous and essential. Chapter 49 describes what it enables: reading and categorizing email, pulling reports, updating dashboards, scheduling and follow-ups, monitoring projects and escalating.

On safety, Chapter 50 says automation done carelessly is dangerous, like a table saw. This is the one place the book tells leaders to bring IT in. Its advice is to hold two instincts at once: move fast on simple connections like Outlook and SharePoint, and slow down on governance, security, training and support.

Chapter 50 is also candid that many companies are still just talking about automation, and that setup is harder than people expect. What to do next
  1. Start with read-only connections to email and documents.
  2. Add write actions one at a time with human approval.
  3. Have IT review permissions and logging before any agent can change records.
 Offering: Implementation
Relevant to: IT and technology, COO and operations, CEO

Sources
What guardrails do AI agents need before they can act?
Short answer. Standing limits on what they can access and change, human approval for decisions needing judgment, logging, and regular review. The book's trip wire is that an agent with standing permission to act and no standing review is a risk. The appendix's fourth step adds guardrails: boundaries that keep the AI within the rules of your business, respecting data sensitivity and decision-making norms. Its example is a financial services firm whose AI retrieved the right data but made confident recommendations in areas needing human judgment and regulatory review. Adding guardrails moved it from decision maker to decision supporter, which the book calls the right design.

The checklist includes roles, document permissions, escalation rules for weak answers, a logging and audit trail, and approval screens. Chapter 17 adds trip wires such as an AI making pricing decisions no human reviews, and a model that was right until conditions changed with nobody watching.

Chapter 27's SOX discussion supports human sign-off wherever a human was always the control. What to do next
  1. List actions the agent may take without approval and those it may not.
  2. Turn on logging for every action.
  3. Review a sample of actions weekly.
 Offering: Implementation
Relevant to: IT and technology, Legal and compliance, CFO and finance, Financial services

Sources
How do we choose which processes to automate first?
Short answer. Start with the easy, repetitive work, and begin at the leadership level where reports, summaries and accountability loops live. Pair automation with a decision layer so you do not get faster wrong calls. Chapter 49 says most organizations automate from the bottom with customer service bots and scheduling tools, which saves time but does not reshape leadership. Real ROI begins when automation starts at the top: meeting summaries, accountability loops, forecasting and performance dashboards. Every major innovation entered the leadership office first.

Chapter 48 warns that automation without a decision layer produces speed with no direction. Chapter 13 gives a cautionary example: Tesla over-automated its Model 3 line, and Elon Musk wrote that excessive automation was a mistake and humans are underrated. He brought robots back more slowly and methodically.

Chapter 22 adds that Air Canada's chatbot, built for a hated task, produced an expensive error, so even easy tasks need review. What to do next
  1. List recurring tasks and score them by hours and risk.
  2. Begin with leadership reporting and meeting follow-ups.
  3. Expand only after the first automation proves stable.
 Offering: Implementation
Relevant to: COO and operations, CEO, Manufacturing

Sources
When should IT be involved in an AI project?
Short answer. Early for security and plumbing, and always before automation can touch systems or records. Not as the owner of strategy. The book is explicit about both halves. The book repeatedly says to keep IT out of the strategic conversation, then Chapter 50 says the opposite for automation: when AI can touch your systems, move your money or change your records, that is IT's job to help secure, so bring them in. Chapter 51 says setting up enterprise accounts, access and connections to data is the single thing IT should be doing.

Chapter 27 also says to talk to your implementation partner about your regulatory environment at the start, since the architecture decisions that determine compliance are foundational. What to do next
  1. Bring IT in at design time for identity, hosting and permissions.
  2. Require IT sign-off on any automation that writes to systems.
  3. Keep use case and data decisions with a business owner.
 Offering: Implementation
Relevant to: IT and technology, CEO

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 27; Ch. 50; Ch. 51
How do we capture and use meeting transcripts in practice?
Short answer. Record and transcribe consistently, tag speakers so weighting works, structure the capture so it adds no friction, and feed transcripts into the shared environment over time. The book says value shows by about six months and compounds past three quarters of data. Chapter 31 describes feeding transcripts as base data and testing them against notes. Chapter 32 adds that digital collection tools like Miro made whiteboard content searchable, taggable and transferable to AI, and that guide rails like values and risk tolerance can be digitized with flexible settings.

Chapter 36 says to flag speakers so the AI knows who is talking and how much weight to give them. Chapter 9 describes a magic unlock once a company has roughly three quarters of accumulated data, closing in on a year. Chapter 14's client had about nine months and a thousand transcripts plus KPIs. Chapter 33 says to keep transcripts flat for now: they carry words, not volume or energy, and that flatness is a feature. What to do next
  1. Choose a recording and transcription tool and a consent policy.
  2. Tag speakers and meeting types.
  3. Feed transcripts in on a schedule and review what the AI surfaces.
 Offering: Implementation
Relevant to: CEO, COO and operations, Sales leader

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 9; Ch. 14; Ch. 31; Ch. 32; Ch. 33; Ch. 36
How do we feed our org chart, job descriptions and business frameworks into AI?
Short answer. Load the real org chart, job descriptions, scorecards, span of control and actual accountability, plus whichever framework you run, such as Scaling Up, EOS or Collins. The book says almost no company has done this, and that it noticeably sharpens results. Chapter 54 says that without an org chart AI is guessing about who reports to whom and how decisions move. It also covers the real structure: who actually talks to whom. Job descriptions, scorecards, what someone believes their job is, span of control and real accountability all belong. The author reports AI warning about a social connection between employees that a sharp human would have caught, and says it will do so once given the chance.

He adds that frameworks, such as Scaling Up, EOS or Jim Collins' work, become their own unlock once inside the AI, and that he has not seen a company do this on its own without being pushed. The workbook has the detailed checklist. What to do next
  1. Collect the current org chart and role scorecards.
  2. Add notes on real decision paths and informal relationships, with care for privacy.
  3. Load your operating framework documents.
 Offering: Implementation
Relevant to: CEO, HR and people, COO and operations

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 54; Ch. 41
How does data weighting work in practice?
Short answer. Tag every source and speaker, then assign numeric weights, for example 1 to 5 or 1 to 10. The book found numbers more reliable than telling the AI in words that one input matters more. Chapter 36 describes the practice at REDEGADES.AI: weights on a 1 to 5 or 1 to 10 scale, with the CEO at the top and a new employee at the bottom. Both scales work, and telling the AI in words that one item is more important was less reliable than numbers. When meetings are recorded, speakers are flagged so the AI knows whose voice it is hearing.

Chapter 46 stresses tagging must be right first: get the tagging wrong and weighting cannot fix it. Chapter 41 says older strategies go in for context but weighted down. Chapter 46 adds that the work is ongoing and describes weighting as a professional skill, noting that few companies do it in any significant way. What to do next
  1. Define the weight scale and who sets it.
  2. Tag all new data with speaker, role and date.
  3. Test with a known disagreement and tune.
 Offering: Implementation
Relevant to: IT and technology, CEO, COO and operations

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 36; Ch. 41; Ch. 46
What is phased ingestion, and what happens in the first 90 days?
Short answer. You teach the AI in stages: cornerstone data first, then voices and context, then operational reality. The book frames it as 90 days, ending with results in the boardroom. Map the company before any of it starts. Chapter 44 describes three phases. Phase 1, Cornerstone Data: strategy documents, core values, organizational design, one- and five-year plans, financial posture and the CEO doctrinal statement. Phase 2, Voices and Context: meeting transcripts, staff interviews, quarterly and annual planning sessions. Phase 3, Operational Reality: sales data, competitive analysis, product details and operational constraints.

The book says to map the company first: structure, doctrine, bias and who owns which decisions. On the calendar, day one is listening and writing down who you are, by day sixty one real use case is live in front of one real executive, and by day ninety you have results. The workbook (Section 2.10) has the week-by-week plan. Rushed ingestion, the book says, produces a system that sounds smart and reasons poorly. What to do next
  1. Map structure, doctrine and decision owners.
  2. Load Phase 1 data and test with the CEO.
  3. Add Phase 2 and Phase 3 data in order.
 Offering: Implementation
Relevant to: CEO, COO and operations

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 44
How do we keep a private AI accurate over time?
Short answer. Plan for upkeep from day one: keep data fresh, run recurring tests for drift and degradation, review prompts, retrieval and access, and revisit governance each quarter. The book says an AI system is not a CRM you configure and forget. Chapter 35 says AI can degrade based on what it is fed or how retrieval was built, and recommends testing like a manufacturer running broken parts through a line to see whether they are detected. It adds that many implementations are built on sand because the technology is young, and to plan for upkeep from the start.

The appendix's optimization stage includes adding data gradually, tuning prompts and retrieval, improving document quality, speed, security and source accuracy, and adding use cases over time. Chapter 18 adds the governance side: AI strategy is not evergreen, so revisit it quarterly against changing regulation and tool use. Chapter 38 notes hallucination rates have fallen but not gone away. What to do next
  1. Schedule monthly test runs and quarterly governance reviews.
  2. Assign an owner for data freshness.
  3. Log changes and results.
 Offering: Implementation
Relevant to: IT and technology, COO and operations, CEO

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 18; Ch. 35; Appendix
How do we build an audit trail for AI-assisted decisions?
Short answer. Capture the prompt, the AI's response, the sources it used and the human decision, in a system you control, with logging enabled and retention rules set. The book calls this what regulators and lawyers will ask for. Chapter 18 explains that most AI use leaves no trail, and that you should be able to show the original prompt and response in sequence and where AI's role ended. The book describes a Decision Alignment Layer that tracks decisions, documents AI's role and builds the audit trail state laws already demand.

The appendix's configuration list includes logging and an audit trail, source citation rules, escalation rules for weak answers, and approval screens where needed. Chapter 8 notes the system can be configured to cite where an answer came from.

The book recommends building governance in now while it is a choice, not a scramble. What to do next
  1. Decide which decisions need a recorded AI trail.
  2. Enable logging and source citation in your AI environment.
  3. Set retention with counsel.
 Offering: Implementation
Relevant to: Legal and compliance, IT and technology, CEO, Financial services, Healthcare

Sources
How do we choose a technical partner for an AI build?
Short answer. The book says do not attempt it solo, and that a partner saves time and costly mistakes. Look for one that starts with your leadership, data and decisions rather than code, and that discusses your regulatory environment before building. The appendix's pro-tip says a technical partner will save a significant amount of time and keep leaders from costly mistakes, and that this is not a solo endeavor. Chapter 27 says to talk to your implementation partner about your regulatory environment at the start. Chapter 20 notes that in rescued projects the failure was usually leadership, not technology.

Chapter 19 describes the author's approach: start with leadership, identify the voices that matter, capture their thinking, structure the data and teach the AI to listen. Ask any partner, including us, the same questions. What to do next
  1. Ask how they capture leadership thinking before building.
  2. Ask how they handle your regulated data and logging.
  3. Ask who maintains the system after launch.
 Offering: Implementation
Relevant to: CEO, IT and technology

Sources
  • Your AI Wake-Up Call by Wade W. Wyant, Ch. 19; Ch. 20; Ch. 27; Appendix
Where do most RAG builds fail?
Short answer. Before the technology is touched. The appendix lists, in order of frequency: nobody owns what data matters, data never gets cleaned, it is run as an IT project, nobody weights the data, testing is skipped, and people build everything at once. The appendix's list is blunt. Nobody owns the decision about what data matters. Raw, duplicated, outdated files get fed in as is. It gets treated as an IT project instead of a business project. Nobody weights the data, so a new hire's guess and the CEO's doctrine count equally. Testing is the first thing cut when deadlines tighten. And teams try to build the whole thing at once rather than proving one real business problem first.

Wyant calls every one a leadership failure, not a technical one: fix ownership, cleaning and weighting before worrying about the vector database, which is the easy part. What to do next
  1. Check your project against all six failure modes.
  2. Fix ownership and data cleaning first.
  3. Schedule testing before the build starts.
 Offering: Implementation
Relevant to: CEO, IT and technology, COO and operations

Sources
How should a COO use AI?
Short answer. Start with the recurring status-chasing and reporting that consumes the week, connect it to operational data, and make sure operational voices are weighted properly in the shared system. The book's examples are status updates, monitoring and escalation. Chapter 49 describes the COO who no longer chases status updates because AI connected to systems monitors projects and escalates when teams get stuck. Chapter 46's plant example shows why weighting matters: a VP of Operations, a three-month-old regional manager and a sales VP disagree about a $200,000-per-quarter extra shift, and unweighted AI hears them about equally.

Chapter 32 adds that digitizing guide rails, such as risk tolerance and geographic approach, with flexible settings, lets the whole team work in the same sandbox. Chapter 13's Tesla story is the caution on over-automating. What to do next
  1. List the three status or reporting loops that cost you the most time.
  2. Connect them to AI with read-only access first.
  3. Add operational constraints and contracts to the shared data.
 Offering: Implementation
Relevant to: COO and operations, Manufacturing

Sources
How can a manufacturer use AI on its own sales and supplier history?
Short answer. Feed historical sales reports, supplier records and demand data into a retrieval system so answers come from what your business actually shows, not what markets usually do. The appendix describes exactly this forecasting example. The appendix's manufacturing CEO wanted better forecasting. Instead of using a generic chatbot, the company pulled in historical sales reports, supplier records and years of demand data. Now answers reference their own history. The book calls that the shift from 'what does the market usually do' to 'what does our business actually show.'

Chapter 9 lists legacy manufacturing among industries REDEGADES.AI has worked in. Chapter 46's night-shift example shows why contracts and operational facts must be in the data, and why weighting matters: a $200,000 per quarter decision was argued by three confident people who had not read the contract. Chapter 13's Tesla lesson cautions against over-automating the line before the human parts are understood. What to do next
  1. Collect sales, supplier and demand history in one place.
  2. Clean and standardize product and customer names.
  3. Start with one forecast question and test it against known outcomes.
 Offering: Implementation
Relevant to: COO and operations, CFO and finance, CEO, Manufacturing

Sources

Sources cited

35 outside sources from the footnotes of Your AI Wake-Up Call. Each lists the answers above that rely on it. Links open the original publication; some publishers may require a subscription.
Industry analysts
  1. Gartner Announces Top Data and Analytics Predictions, Gartner, June 17, 2025
    Used in: Why should the CEO use AI first?; How do we train staff to use AI?
  2. AI Agent Layer, Gartner
    Used in: What is the Decision Alignment Layer?; What are AI agents and MCPs, and how do we automate safely?; What is IT's role in an AI project?; Where does AI ROI actually show up in the P&L?
  3. Transformational Moment: Decision Intelligence and the 2025 Gartner AI Hype Cycle, Aera Technology
    Used in: What is the Decision Alignment Layer?; Where does AI ROI actually show up in the P&L?
  4. Gartner: AI Agents Will Drive Half of Decisions by 2027, Technology Magazine
    Used in: What is the Decision Alignment Layer?; What are AI agents and MCPs, and how do we automate safely?; Where does AI ROI actually show up in the P&L?
  5. Enterprise AI adoption survey results, WRITER
    Used in: What is a shared company brain, and why does it matter?; Why do our leaders disagree more since we adopted AI?
  6. Gartner Says More Than Half of Enterprise IT Spending in Key Market Segments Will Shift to the Cloud by 2025, Gartner, February 9, 2022
    Used in: How do we avoid AI vendor lock-in?
 Research and reporting
  1. Why 95% Of AI Pilots Fail, And What Business Leaders Should Do Instead, Andrea Hill, Forbes, August 21, 2025
    Used in: Why do most AI initiatives fail?; Where do most RAG builds fail?
  2. MIT Report Finds Most AI Business Investments Fail, Reveals 'GenAI Divide', Virtualization Review, August 19, 2025
    Used in: Why do most AI initiatives fail?; Where do most RAG builds fail?
  3. What Are AI Hallucinations?, IBM
    Used in: What is an AI hallucination and how should staff handle it?; How do we test an AI system for hallucination and degradation?; Why does AI give confident but wrong answers on our own data?
  4. Study finds nearly two-thirds of AI-generated citations are fabricated or contain errors, PsyPost
    Used in: How do we check AI output before it goes to a client or the board?; What is an AI hallucination and how should staff handle it?; How do we test an AI system for hallucination and degradation?
  5. Zillow to shutter home-buying business and lay off 2,000 employees as its big real estate bet falters, GeekWire, November 2, 2021
    Used in: How do we check AI output before it goes to a client or the board?; How should a financial services firm use AI without handing it decisions?
  6. Flip Flop: Why Zillow's Algorithmic Home Buying Venture Imploded, Stanford Graduate School of Business
    Used in: How do we check AI output before it goes to a client or the board?; How should a financial services firm use AI without handing it decisions?
  7. Elon Musk says humans are underrated after Tesla robot production issues, Global News, April 2018
    Used in: How do we choose which processes to automate first?; How should a COO use AI?; How can a manufacturer use AI on its own sales and supplier history?
  8. IBM sells Watson Health assets (paywalled), STAT News, January 21, 2022
    Used in: How can a healthcare organization use AI without violating HIPAA?
  9. Air Canada chatbot costs airline discount it wrongly offered customer, CBS News, February 2024
    Used in: What guardrails do AI agents need before they can act?; Why do leaders automate the wrong things first?; What is data weighting and why does it matter?
  10. How AI can make us more receptive to negative feedback, Cambridge Judge Business School, 2024
    Used in: How do we use AI as a thought partner in meetings, not just a search tool?; How do we get AI to challenge us instead of just agreeing with us?
  11. Zuckerberg is training an AI agent to do CEO work, Futurism
    Used in: What is a shared company brain, and why does it matter?
  12. Cloudflare outage briefly takes ChatGPT, Claude and other services offline, SiliconANGLE, November 18, 2025
    Used in: How do we avoid AI vendor lock-in?; Is a multi-LLM setup worth it for a mid-sized company?
  13. Microsoft shuts down AI chatbot after it turned into a racist, CBS News, March 2016
    Used in: What is business bias and why would we teach it to AI?
  14. Lionel Messi and the long arc of a soccer superstar, Yahoo Sports
    Used in: Why do employees learn AI better from a person than from an AI tutor?
  15. Post introducing "vibe coding", Andrej Karpathy on X, February 2025
    Used in: What is shadow AI and how do we get it under control?
  16. Why AI Will Save the World, Marc Andreessen, a16z, 2023
    Used in: What is shadow AI and how do we get it under control?
  17. McDonald's will reportedly keep dozens of items off the menu, CNBC, June 18, 2020
    Used in: How do we choose which processes to automate first?; How should a COO use AI?
  18. OpenAI o1 self-preservation behavior, Futurism
    Used in: What guardrails do AI agents need before they can act?
 Law and regulation
  1. Article 86: Right to Explanation of Individual Decision-Making, EU Artificial Intelligence Act
    Used in: How should a financial services firm use AI without handing it decisions?; What do the EU AI Act and US state AI laws mean for a mid-sized company?; If we were served a subpoena for our employees' AI conversations, could we produce them?; What does AI governance look like for a 50-person firm?; What should general counsel ask about AI use in the company?; What should HR know before using AI in hiring and performance reviews?; How do we build an audit trail for AI-assisted decisions?
  2. EU Approves Delays and Other Amendments to Certain EU AI Act Obligations, Morgan Lewis, June 2026
    Used in: What do the EU AI Act and US state AI laws mean for a mid-sized company?; If we were served a subpoena for our employees' AI conversations, could we produce them?; What should general counsel ask about AI use in the company?
  3. State AI legislation tracker, multistate.ai
    Used in: What do the EU AI Act and US state AI laws mean for a mid-sized company?; What does AI governance look like for a 50-person firm?
  4. AI hiring laws by state: compliance map, AI Laws by State
    Used in: What do the EU AI Act and US state AI laws mean for a mid-sized company?; What should HR know before using AI in hiring and performance reviews?
  5. Colorado Rewrites Its Landmark AI Law: Unpacking SB 26-189, Consumer Finance Monitor (Ballard Spahr), May 12, 2026
    Used in: What do the EU AI Act and US state AI laws mean for a mid-sized company?; What does AI governance look like for a 50-person firm?; How do we build an audit trail for AI-assisted decisions?
  6. HIPAA Compliance for AI in Healthcare, Medcurity
    Used in: We're in a regulated industry. Is compliance a reason to wait on AI?; What do HIPAA and SOX actually require when we use AI?; How do we use AI on our own documents without exposing data?; How can a healthcare organization use AI without violating HIPAA?; What should general counsel ask about AI use in the company?
  7. Strengthen SOX Compliance: Assessing Risk Materiality of AI Enablement, Schneider Downs
    Used in: We're in a regulated industry. Is compliance a reason to wait on AI?; What do HIPAA and SOX actually require when we use AI?; How do we use AI on our own documents without exposing data?; How should a financial services firm use AI without handing it decisions?; How should a CFO use AI?; What should general counsel ask about AI use in the company?
 Books
  1. Nine Lies About Work, Marcus Buckingham and Ashley Goodall, Harvard Business Review Press, 2019
    Used in: What should HR know before using AI in hiring and performance reviews?; Why do leaders automate the wrong things first?; How should a sales leader use AI with call transcripts and KPIs?
  2. Predictable Success, Les McKeown
    Used in: How can an owner-led company grow past 25 employees with AI?
  3. Built to Last, James C. Collins and Jerry I. Porras
    Used in: How do we get AI to challenge us instead of just agreeing with us?
  4. Winning, Jack Welch and Suzy Welch
    Used in: How do we get AI to challenge us instead of just agreeing with us?

About this library

The answers summarize and cite Your AI Wake-Up Call by Wade W. Wyant (REDEGADES.AI, 2026). Statistics and case studies are the book's own and are attributed to it and to the outside sources listed above. Where an answer applies the book to a specific role or industry, that is stated in the answer. Nothing here is legal, compliance or financial advice. Confirm regulatory questions with qualified counsel. Learn more about our offerings: Training & Enablement, Design & Strategy and Implementation.

Not sure which question is yours?

Book a short assessment and we will point you to the right starting place for your team. Book An Assessment