Mulally Clapped, and Ford Stopped Dying

What Ford’s CEO understood about culture is now the difference between companies that survive AI and companies that quietly fracture.

In 2006, Ford was on track to lose $17 billion. The company was paralyzed by a culture of fear, secrecy, and internal competition. Showing weakness in an executive meeting was historically a fireable offense.

Alan Mulally arrived as CEO and instituted a single ritual: a weekly Business Plan Review. Every executive had to color-code their progress. Green for on track. Yellow for at risk. Red for off-plan with no solution.

For weeks, despite the company hemorrhaging billions, every chart came back green.

Then Mark Fields, head of the Americas, showed a red slide for a major launch delay on the Ford Edge. The room froze. Everyone expected Fields to be terminated on the spot.

Mulally clapped.

He praised Fields for the visibility and asked the room, “Who can help Mark with this?”

That single moment broke Ford’s fragile culture. Executives stopped hiding their failures and started pooling their expertise to solve them. Ford avoided the bailout that humbled its competitors and returned to profitability.¹

What Mulally built at Ford was organizational resilience, the ability for a company to absorb shocks, learn from failures, and adapt faster than the environment is changing. It is the same characteristic every American company now needs to develop, because AI has accelerated innovation and transformation.

Industry disruption is occurring weekly. Companies with a fragile culture that has to assign blame in order to learn cannot learn fast enough to compete.

This article gives you the framework to understand what is about to happen to your team, a tool to measure where you stand, and a plan for what to do Monday morning.

What AI Is Actually Going to Do to Your Team

When a company adopts AI in a meaningful way, three things change at once.

1. The tools your team uses change, sometimes every few weeks.

2. The work itself changes, because tasks that took hours now take seconds and tasks that used to be impossible become routine.

3. The question of who does what changes, because some of the work will be done by software, while new kinds of work will appear that no one on your team has done before.

That is the test your company is now facing. The deciding factor is whether your culture can match the pace of adaptation the technology requires.

Resilience Is a Set of Behaviors. Act Accordingly.

When most leaders hear the word resilience, they think of grit, perseverance, or a positive attitude. The research on how organizations actually survive crisis tells a different story. Resilience is something teams do, made up of specific behaviors you can watch happening on any given Tuesday afternoon.

Two researchers, Karl Weick and Kathleen Sutcliffe, spent decades studying organizations that operate in genuinely high-stakes environments and almost never fail catastrophically. Nuclear aircraft carriers. Air traffic control towers. Hospital emergency rooms. They wanted to understand why these organizations, despite handling situations where one mistake can have fatal consequences, have remarkably few catastrophic failures.

What they found is that resilient teams share a set of habits they called mindful organizing.

• The teams pay attention to small problems before they become big ones.

• They share information openly.

• They treat the person closest to a problem as the expert on it, regardless of rank.

• They assume failure is always possible and design their conversations accordingly.²

Fragile teams do the opposite.

• They suppress small problems hoping those problems will resolve themselves.

• They report progress in ways that protect their reputations rather than reflect reality.

• They route every decision through hierarchy.

• They assume that past success means future success, and stop watching for the failure modes they have not yet experienced.

The difference between the two shows up in four places.

There is a phrase people inside fragile companies use to describe the way bad news gets hidden. They call it watermelon reporting 🍉. Green ✅ on the outside, red ❌ on the inside.

The status report says everything is on track. The reality is that the project is failing, and everyone close to it knows. The leadership above does not know, because the culture has trained the people below them that surfacing the truth is dangerous.

In resilient organizations, the working assumption is different.

When something goes wrong, leadership starts by asking what conditions made the mistake reasonable from the perspective of the person who made it. Failures are treated as evidence that the system has a flaw, not that a person has a flaw. That single shift in assumption is what makes telling the truth rational instead of risky.³

The same logic shows up at Etsy, the online marketplace. Under engineering leader John Allspaw, Etsy deployed new code to its live website dozens of times every day. That pace would be impossible in a company where engineers feared being blamed for outages. Allspaw made a deal with his team. If an engineer who caused a problem explained exactly what they did and why it seemed like the right call at the time, the company would not punish them. In return, the engineer was responsible for helping the company learn from the mistake so it would not happen again. As Allspaw put it, engineers were on the hook for making the system safer.⁴

Blamelessness makes high standards possible.

Why AI Makes This Test More Urgent

For most of business history, the changes that hit a company arrived slowly enough that even fragile cultures could survive them. A bad year was followed by a good year. Mistakes had time to surface. The cost of hiding bad news was real but manageable, because there was usually time to recover from the mess hidden bad news creates.

AI changes this in a specific way: it removes the time.

Here is what that looks like in practice. A team adopts an AI tool to handle customer service inquiries. The tool works well at first. Three months in, an employee notices that the tool is occasionally giving customers incorrect information about return policies.

In a healthy culture, the employee mentions it in the next team meeting, the team investigates, and they fix the underlying issue before any customer is harmed. The fragile version of this story ends with a regulatory complaint, a class-action lawsuit, or a viral social media incident, all tracing back to the moment one employee noticed something was off and decided not to say.

That gap between noticing and saying is what fragile cultures produce. AI makes that gap catastrophic, because the speed at which AI tools change means that a problem hidden today will compound into a much larger problem within months, not years.

The research backs this up.

A 2025 study in the journal Strategy and Leadership looked at companies adopting AI and found something important: the technology by itself does not make a company more resilient. What matters is whether the company already has the cultural ingredients in place, specifically a culture of innovation, an ability to move quickly, and leadership that understands the change. Companies with those ingredients found that AI made them stronger. Companies without them found that AI exposed every weakness their culture already had.⁵

A separate study found that when employees feel anxious about whether AI will eliminate their jobs, they start hiding what they know from their coworkers, because they believe their unique knowledge is what protects their position. That hoarding behavior makes the team worse at solving problems exactly when the team needs to be solving problems faster.⁶

The pattern is consistent.

Healthy cultures absorb AI and grow. Unhealthy cultures absorb AI and crack. The technology amplifies whatever was already there.

There is one more signal worth paying attention to inside your own company. Notice who gets celebrated. In fragile organizations, the heroes are the people who work eighty-hour weeks to put out fires. In resilient organizations, the heroes are the people who notice systemic flaws before they become fires.

Whichever your company actually rewards tells you which kind of company you actually run, regardless of what the values poster says.

A Tool for Measuring Your Team’s Readiness

So how do you know which kind of culture your team has? The good news is that researchers have already built a tool for exactly this question.

It is called the Mindful Organizing Scale, developed by Timothy Vogus and Kathleen Sutcliffe (the same Sutcliffe who studied high-reliability organizations).⁷ It is a short, validated assessment that measures the specific behaviors that separate resilient teams from fragile ones. The scale has been tested across high-stakes environments like hospitals and traditional enterprise workplaces, and it consistently produces reliable results.

Here is what makes it useful for the AI moment we are in. The scale was originally built to measure how teams handle high-stakes complexity, the kind of environment where small failures can cascade quickly. AI integration creates exactly that kind of environment inside ordinary companies. The scale is a nine-question instrument, and each question turns out to map directly onto a specific way that AI breaks fragile organizations.

Take the Assessment

Rate each question from 1 to 5, where 1 means strongly disagree and 5 means strongly agree. Answer based on how your immediate team actually operates, not how you wish it operated. The honest answer is the only useful answer.

1. When new problems come up, we talk with coworkers about what to watch for.

2. We spend time naming the work we cannot afford to get wrong.

3. We talk about different ways to do our normal work.

4. We know what each person on the team is good at.

5. We share our specialized skills so everyone knows who to go to.

6. In a crisis, we quickly combine what we each know to solve it.

7. We talk about mistakes and what they teach us.

8. When something goes wrong, we talk about how to prevent it next time.

9. When someone raises a problem, we look for a bigger pattern instead of treating it as a one-off.

Add up your scores. The total will fall between 9 and 45.

38 to 45: Highly Resilient.

Your team treats anomalies as useful information, defers to whoever has the relevant expertise, and views failures as opportunities to improve the system. You are positioned to absorb AI disruption and turn it into an advantage.

28 to 37: Transitional.

Your team has some resilient habits but falls back on hierarchy or blame when under real pressure. You may have the agility to react to AI-driven change, but you lack the safety to learn from it as fast as the technology requires.

9 to 27: Fragile.

Communication is siloed, bad news is suppressed, and failure is treated as an individual flaw. The pace of AI-driven change will cause significant fracture and burnout in your team within the next year or two unless leadership intervenes now.

The Real Diagnostic Is the Gap Between Scores

Your individual score matters. What matters more is what happens when every member of your team takes the assessment independently and you compare the results.

If individual scores vary by more than eight to ten points across your team, that gap is the real signal. It means psychological safety is not distributed equally. Almost always, leaders score the team higher than the frontline does. The leaders see a team that talks about problems openly. The frontline sees a team that has learned to perform openness in front of leadership and tell the truth elsewhere.

That gap is exactly what AI disruption exploits. The bad news is happening. The signals are arriving. The frontline sees them. Leadership does not, because the culture has trained the frontline that surfacing them is unsafe.

A fragile culture in 2006 was a slow-motion problem. A fragile culture in 2026 is an acute one.⁸

What to Do Monday Morning

Send the assessment to every member of your team. Have them complete it independently. Aggregate the results without identifying individuals. Compare your own score to the team’s average.

If the gap is meaningful, that is the conversation that matters more than the score.

Then bring the team together and use these three prompts to debrief.

Think of the last time a major project failed or missed its target. Was the immediate focus on who caused it, or what systemic factors allowed it?

If an entry-level employee spotted a flaw in an AI implementation plan designed by a vice president, how comfortable would they be raising the issue in a team meeting?

What is the unwritten rule on our team about delivering bad news to leadership?

The answers will tell you what Mulally figured out the week he started clapping for red slides. The information you need is already inside the building. The only question is whether you have built a culture where it can reach you in time.

Subscribe now

Appendix: How Each Question Predicts an AI-Specific Failure Mode

Walking through the questions one at a time shows why this assessment is the right tool for this moment.

Question 1: When discussing emerging problems with co-workers, we usually discuss what to look out for.

AI tools produce warning signs before they produce outright failures. A customer service AI starts giving subtly wrong answers in unusual cases. An automated decision tool begins making choices that drift outside what it was designed for. A workflow tool quietly degrades the quality of an output without anyone noticing for weeks.

Teams that habitually talk about what to watch for catch these warning signs early. Teams that do not catch them only after the failures have stacked up.

Question 2: We spend time identifying activities we do not want to go wrong.

The most expensive AI failures happen in places where no one was watching, because no one named those places as critical in advance. Imagine a hospital that deploys AI to help with patient scheduling. If the team identified billing accuracy as a critical activity that cannot fail, they will check the AI’s effect on billing carefully. If they did not name it as critical, they will discover the billing problem from an angry insurance company months later.

Teams that map out what cannot fail before they deploy AI catch problems quickly. Teams that deploy AI broadly without that mapping discover their critical paths only after those paths have already broken.

Question 3: We discuss alternatives as to how to go about our normal work activities.

This is the question that most directly predicts how a team will handle AI integration. A team that rigidly defines work as a single fixed process cannot absorb AI-driven workflow change. When the new tool changes how the work gets done, the team experiences it as disruption rather than adaptation.

A team that habitually discusses alternative ways to do its work has already built the cognitive flexibility AI integration requires. The AI transition feels like a continuation of how the team already operates.

Question 4: We have a good map of each person’s talents and skills.

AI redistributes work based on what humans add that machines cannot. Some routine tasks your senior people do today will be done better by AI within a year. Some work that requires judgment, context, or creativity will become more important.

If you know your people deeply, you can reassign work intelligently as the technology absorbs the routine. If you operate on title and seniority alone, you end up with senior people doing work AI has now made obsolete, while junior people with the relevant judgment go underused.

Question 5: We discuss our unique skills with each other so that we know who has relevant specialized skills.

When an AI deployment surfaces a problem no one anticipated, someone on the team usually has the specific knowledge to solve it. Whether that knowledge gets to the problem in time depends on whether the team already knows who has it.

Teams that have made their expertise visible to each other route problems to the right person quickly. Teams that have not waste critical hours discovering capabilities they already had.

Question 6: When a crisis occurs, we rapidly pool our collective expertise to attempt to resolve it.

The defining feature of AI disruption is speed. The pace of new model releases, capability shifts, and downstream changes to how work gets done outruns the response capacity of slow-moving organizations.

Teams that pool expertise quickly survive the speed. Teams that route every decision up through hierarchy find that their decisions arrive after the window for them has already closed.

Question 7: We talk about mistakes and ways to learn from them.

Early AI deployments fail often. The error rate of new tools, new workflows, and new handoffs between humans and AI is high by design, because the systems are still being calibrated to the specific company. This is normal and expected.

Teams that openly discuss mistakes turn each failure into useful information that improves the next iteration. Teams that hide mistakes accumulate hidden problems that compound until something serious breaks.

Question 8: When errors happen, we discuss how we could have prevented them.

This question separates teams that learn forward from teams that learn defensively. Forward learning produces actual changes to how the work gets done. Defensive learning produces blame and the appearance of accountability without any real change.

AI accelerates this distinction because the same kinds of errors repeat across deployments. A team that examines prevention systematically gets better with every error. A team that just assigns fault keeps making the same kinds of errors with each new tool.

Question 9: When someone brings up a problem, we look for a larger pattern rather than dismissing it as a one-off event.

This is the highest-leverage question in the AI context. AI problems almost never arrive as isolated events. A single bad output is usually a symptom of something larger: a misaligned prompt, a flawed integration with an existing system, an issue with the data the AI was trained on, or a workflow assumption that no longer holds in the new environment.

Teams that look for the pattern catch the underlying problem early. Teams that dismiss the report as a one-off keep paying for the same hidden failure across dozens of downstream incidents.

The nine questions, taken together, measure whether a team can sense, route, learn from, and adapt to fast-moving change. AI is the highest-frequency change most companies have ever faced. A team that scores high on this instrument is built for exactly the kind of pressure AI creates. A team that scores low is not. The score does not tell you whether your team can adopt the technology. It tells you whether your team can survive the integration.

Appendix: References

¹ Hoffman, Bryce G. American Icon: Alan Mulally and the Fight to Save Ford Motor Company. New York: Crown Business, 2012.

² Weick, Karl E., and Kathleen M. Sutcliffe. Managing the Unexpected: Sustained Performance in a Complex World. 3rd ed. Hoboken, NJ: John Wiley & Sons, 2015.

³ Dekker, Sidney. The Field Guide to Understanding ‘Human Error’. 3rd ed. Farnham, UK: Ashgate Publishing, 2014.

⁴ Allspaw, John. “Blameless PostMortems and a Just Culture.” Code as Craft (Etsy engineering blog), May 22, 2012. https://www.etsy.com/codeascraft/blameless-postmortems.

⁵ “Artificial Intelligence and Organizational Resilience: The Mediating Role of Agility, Innovation, and Digital Leadership.” Strategy & Leadership. Emerald Publishing, 2025. https://www.emerald.com/sl/article/doi/10.1108/SL-08-2025-0275.

⁶ “How Artificial Intelligence-Induced Job Insecurity Shapes Knowledge Dynamics: The Mitigating Role of Artificial Intelligence Self-Efficacy.” Journal of Innovation & Knowledge, 2024. https://www.sciencedirect.com/science/article/pii/S2444569X2400129X.

⁷ Vogus, Timothy J., and Kathleen M. Sutcliffe. “The Safety Organizing Scale: Development and Validation of a Behavioral Measure of Safety Culture in Hospital Nursing Units.” Medical Care 45, no. 1 (2007): 46–54.

⁸ Edmondson, Amy C. The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth. Hoboken, NJ: John Wiley & Sons, 2018.

Additional Background Reading

Roberto, Michael A. Why Great Leaders Don’t Take Yes for an Answer: Managing for Conflict and Consensus. 2nd ed. Upper Saddle River, NJ: Pearson FT Press, 2013.

Taleb, Nassim Nicholas. Antifragile: Things That Gain from Disorder. New York: Random House, 2012.

Wildavsky, Aaron. Searching for Safety. New Brunswick, NJ: Transaction Publishers, 1988.

A note on sources. The Mulally narrative draws on Bryce Hoffman’s reporting from inside Ford during the turnaround. Hoffman covered Ford for the Detroit News from 2005 onward and was granted unprecedented access to Mulally, Bill Ford, the Ford family, senior executives, and internal company documents during the writing of American Icon. The story has been retold in many secondary sources, but the specific dialogue and scene details trace back to Hoffman’s primary reporting.

AI evolves weekly. Every week, there’s an announcement for a new model, disrupted industry, or automated job category.

Businesses evolve over years. They plan quarters ahead for innovation and transformation.

Business that do not accelerate their evolution to keep pace with AI will die.

The Engine That Powers Evolution

Businesses evolve by learning. They cannot evolve faster than they can learn. Learning is the rate-limiting input. If learning slows, evolution slows, and competitive advantage erodes. A company’s learning velocity is therefore the leading indicator of its future competitive position.

Currently, businesses learn how to do new things primarily through two activities:

• Innovation = Go from zero to one, from nothing to something.

• Transformation = Restructure what the business does and how it does it.

Both of these processes imply a start and end date. They are inherently time bound endeavors.

AI’s evolution is unbounded. It will evolve until it automates many industries. Therefore, businesses need to learn to continuously evolve too.

Business Evolution is the process by which the characteristics of businesses change over time, new types of businesses develop, and others disappear. This has always happened; now it will happen at hyper-speed.

Why Evolution Is the Only Durable Advantage

The company that evolves fastest wins, because every other advantage decays. There Is No Steady State

AI’s frontier is moving forward far and fast. It collapses the cost and time of completing tasks, capabilities, products, and processes. It commoditizes software, so it is easier and faster to build things than to buy them.

Either a company learns to evolve, or it stagnates. There is no third option. In a changing environment, standing still is moving backward relative to those powered by AI.

How Evolution Stays Productive Rather Than Random

The operational discipline that keeps evolution from becoming chaos is Amazon’s working-backwards model: start with the customer and work backwards using data.

Without a customer anchor, evolution becomes thrash. The customer is the fixed point. Data is the navigation tool. Every evolution decision starts with a real customer need and reasons backward to what the business must build, change, or scale. As AI makes it cheap to build the wrong thing fast, this discipline matters more than ever.

Why Most Companies Cannot Do This

Now we need to look at why this is so hard for most companies, and what they have to change.

The classic Rogers diffusion curve segments any population into five groups by their relationship to new ideas:

• Innovators (2.5%) create the new

• Early Adopters (13.5%) take risks on the new

• Early Majority (34%) adopt once it is proven

• Late Majority (34%) adopt once it is standard

• Laggards (16%) adopt only when forced

This curve is usually used to describe customers. But it also describes companies. Whole organizations sit somewhere on this curve based on their relationship to change.

For chance, we bucket the curve into three behavioral zones:

• Change Frequently, where Innovators and Early Adopters live

• Change Infrequently, where the Early and Late Majority live

• Rarely Change, where Laggards live

Over time, mature businesses naturally drift right. It is normal for a company that has scaled, optimized, and stabilized to find itself in the Change Infrequently or Rarely Change zone. Stability is the reward for past success, and stability resists change.

The further right a company sits, the more allergic to change its culture becomes, and the less capable it is of evolution. Companies on the left enjoy change. Companies in the middle tolerate it with discomfort. Companies on the right hate it.

AI is now the most aggressive force pulling on every company’s position. Companies on the left absorb AI as augmentation. Companies on the right experience it as existential threat.

What Determines Where a Company Sits

Five cultural attributes determine a company’s position on this curve and its capacity to evolve in the AI era. These are durable cultural drifts; they existed before AI and will exist after. Maturity pulls them rightward by default; the work is to reverse the drift.

Five Cultural Drifts to Reverse

Each is a spectrum. The left pole is the culture of a company that evolves. The right pole is the culture of a company that stagnates:

1. Resilience to Fragility

2. Growth Mindset to Fixed Mindset

3. Continuous Learning to Stagnation

4. Data-Driven Culture to Siloed, Political, and Tribal Culture

5. Customer Obsession to Customer Apathy

As a business matures, time itself pulls the culture rightward on all five dimensions simultaneously. This is the cultural drift. It happens by default. It does not require bad leadership or bad people. It requires only the passage of time and the natural human preference for predictability over disruption.

The Job of Leadership

The job of leadership is to reverse the drift on all five universal dimensions to allow their business to thrive.

A growth mindset without continuous learning is just optimism. Customer obsession without data is just opinion. Resilience without a growth mindset is just stubbornness. The five must move together, deliberately, against the gravitational pull of organizational maturity and against the disruption of an AI-saturated workplace.

The Five in Detail

1. From Fragility to Resilience

Resilience is the capacity to absorb shocks, recover from failure, and keep moving. Fragility is the opposite: small disturbances cause large damage, and recovery is slow or impossible. A company that cannot absorb the failure of an experiment cannot run experiments. A company that cannot run experiments cannot innovate. In the AI era, capability releases, workflow shifts, and identity disruption arrive at high frequency. Resilient cultures absorb these waves; fragile ones fracture.

Resilience is the foundation. The other dimensions sit on top of it. Without resilience, no other cultural change holds.

2. From Fixed Mindset to Growth Mindset

A growth mindset is the belief that abilities and intelligence can be developed through effort, learning, and persistence. A fixed mindset is the belief that they are static traits that cannot be significantly changed.

Reversing toward growth mindset requires balancing short-term quarterly delivery with long-term strategic delivery, emphasizing exploration and experimentation over efficiency and optimization, eliminating organizational debt in the form of rigid processes and norms, and shifting leadership focus from stability and execution to growth and learning.

AI raises the stakes: it continuously redefines the skills the company needs, and a fixed-mindset culture treats that redefinition as a personal threat rather than a learning opportunity.

3. From Stagnation to Continuous Learning

Continuous learning is an ongoing commitment to acquiring new knowledge, skills, and insights, and applying them to improve products, processes, and outcomes. Stagnation is a lack of growth or improvement over time, often due to a failure to learn, adapt, or evolve.

Reversing toward continuous learning requires making learning a strategic priority rather than an HR program, balancing exploitation of current capabilities with exploration of new ones, replacing conformity and risk avoidance with curiosity and experimentation, and replacing siloed and political knowledge with shared and collaborative knowledge.

AI raises both the floor and the ceiling: stagnant organizations fall further behind every quarter, while learning organizations compound their lead.

4. From Siloed, Political, and Tribal Culture to Data-Driven Culture

A data-driven culture values and relies on empirical evidence, metrics, and analytics to inform choices, strategies, and operations at all levels. A siloed, political, and tribal culture relies on localized, subjective, politically influenced knowledge, opinions, and experiences to inform decisions within isolated groups.

Reversing toward a data-driven culture requires eliminating competing agendas built on personal and tribal interests, distributing customer and performance data widely so legacy beliefs can be tested against reality, mandating curiosity and collaboration rather than tolerating silos, and simplifying organizational complexity to reduce communication and trust barriers.

AI is fundamentally a data system: data-driven cultures unlock its full value, while political cultures use AI to confirm preconceptions and ignore inconvenient outputs.

5. From Customer Apathy to Customer Obsession

Customer obsession means the company is organized around the customer at every level. Customer apathy means the company is organized around itself, its politics, and its internal priorities, with the customer as an afterthought.

Reversing toward customer obsession requires shrinking the distance between the organization and its customers, aligning internal priorities and metrics on customer outcomes, emphasizing customer research, feedback, and co-creation, and shifting culture from efficiency and standardization to empathy and service.

AI deepens this focus: customer-obsessed cultures will use AI to anticipate and serve at unprecedented depth and personalization.

The Through-Line

Business success is evolution. Evolution is innovation plus transformation, anchored to the customer and powered by learning. Every business that achieves scale begins to drift culturally to the right on five dimensions, becoming more fragile, more fixed, more stagnant, more siloed, and more inwardly focused. This drift is the default. It is the price of maturity.

The job of leadership is to actively reverse the drift on the five. Without these five, the business stagnates and AI becomes a procurement line item that never produces returns.

With them, the business evolves and AI becomes the most powerful augmentation tool in the company’s history.

What AI Asks of Every Company

Evolve or die.

If you want your business to evolve, follow me.

Subscribe now

With AI, most businesses are twenty years behind the present. Thankfully, AI has made it easier than ever for everyone to evolve their business, catch up, and get ahead.

If the business I’m about to describe sounds far fetched, it’s not. This is how Amazon ran when I worked there in 2007: pre-AI.

The Business That Already Lives Here

It is Tuesday morning. Suzy, a mid-market CEO opens her laptop and reads three things before her coffee is cool.

What happened overnight.

• Pricing adjusted on eleven SKUs after a competitor move hit the data at 2 a.m.

• Inventory rebalanced across three DCs as weather shifted demand in the Southeast.

• A support subroutine resolved four hundred tickets without a human, and flagged six that needed one.

She did not approve any of this. She set the goals, guardrails, and escalation conditions. The loops did the rest.

What opportunities opened up.

For her morning read, her AI scanned all emails, chats, and inbound documents to identify three opportunities that align with her current priorities:

1. A new channel competitors have not touched.

2. A technology shift that compresses her supply chain by a week.

3. A customer segment signaling demand for a product she does not yet sell.

Each comes with recommended next steps, a clear owner, and potential questions. She approves two to move to the next steps and asks the third for more information.

What her team is working on, and the inputs driving it.

She has real time visibility into the inputs her team is actually using: selection, price, speed, quality, experience - all powered by technology with excellent telemetry. These are the levers that cause scalable outputs in business. Outputs are growth in revenue, margin, and retention and reduction in costs.

Where the loops are running well, she lets them run. When they are not, she can read it in the inputs and intervene before it impacts the outputs.

Her competitors are still waiting for the PowerPoint with the monthly numbers.

They are still debating output metrics they cannot clearly define, let alone control. They are running nineteenth century management techniques at a twentieth century slide deck rhythm against a twenty first century atomic clock.

It is 2026. They did not jump on machine learning in 2010. They woke up to AI when ChatGPT launched. They are so far behind that she does not have to worry about them. She keeps her eyes on her customer and on the horizon, watching for her actual competition: the AI native companies that will endure and thrive.

This is not a vision of 2040. It is how a growing number of businesses operate right now. It is also, not coincidentally, how Amazon operated in 2007.

Explaining Business to Computers

To run your business like Suzy, you need to be able to explain it to a computer. To make computers understand business, we have to simplify it for them.

I’ve spent the better part of three decades helping computers operate businesses in many sectors: law, investing, books, ecommerce, payments, physical retail, digital media, advertising, marketplaces, education, and web services.

What I’ve learned from building technology that runs businesses.

Businesses sell solutions.

Solutions are made up of some or part of brands, design, distribution, experiences, management, manufacturing, products, services, technology, and many other potential inclusions.

To build and change solutions, businesses use a few tools:

• Innovation is the creation of new solutions. Going from zero to one.

• Iteration is making a solution incrementally better, going from version 1 to version 2, 3, 4, and onward.

• Transformation is overhauling a solution to make it fundamentally perform differently, ideally making the solution measurably more efficient and effective than before.

These activities were always thought of as projects.

• They have a start date and an end date.

• They are scheduled months in advance through rigorous, continuous planning cycles and project management.

• Budgets were allocated for the discrete, time-bound activities.

The decisions we make, and the data we use to make those decisions, are more critical than ever.

AI changes all of this.

AI changes faster than any technology in history. It is being adapted and adopted faster than any technology in history. And there is more money being invested in AI capabilities than into any other single technology in the history of our species. Things are very different, right now.

It is time to adapt. That starts with learning where you are. The better we understand our own businesses, the more precisely we can start to build the cybernetic loops we need in our business to ensure that we are thriving in the era of AI.

The decisions we make, and the data we use to make those decisions, are more critical than ever. We must learn to encode them into our own cybernetic loops.

This is how we will maximize the application of AI in our job, company, and industries.

The Cybernetic Loop

This is not a metaphor. This is the actual structure of what your business does.

Note: The cybernetic control system serves to keep the system between acceptable operating limits (ex. constraints, performance levels, etc).

Every subroutine in your business runs on the same loop.

1. Target = Your goal

2. Sensor = Your Current Performance

3. Difference = Gap between Our Goal and Performance

4. Controller = Your Options, choose one or many.

5. Action = Execute the Chosen Option(s)

6. System Under Control = The solution you’re changing

7. Change = The outcome of your inputs

8. Feedback = Measure Results from the Action

This is a continuous loop. The speed at which the loop closes and repeats is the variable that determines whether you pull ahead or fall behind.

Cybernetic Control System & Decision Frameworks

We use the cybernetic loop as our foundational process as it aligns with all successful business improvement and innovation decision frameworks. Every framework you have ever used is a variant of the same loop.

You already know this pattern. You have just been running it slowly and in pieces.

Table Source: Pereira, Steve, and Andrew Davis. Flow Engineering: From Value Stream Mapping to Effective Action. IT Revolution Press, 2024, p. 27.

Every framework in this table was designed for loops that close in weeks or quarters. AI closes them in seconds. The frameworks are not wrong. The manual processes underneath them are. They were built for a physics that no longer applies.

The work ahead for most businesses is to accelerate these loops as fast as possible. The rest of this piece and the next several articles I am sharing will show you how.

Cybernetics and Business Management

The more precisely you understand your subroutines and the more specifically you can name the impact each one has on your customer, your sales, your profit, and your growth... the more you are connecting your inputs to your outputs.

This is when your judgment compounds. You begin to trust which investments will produce which outcomes. From there, you can pursue the priorities that matter most at scale, over time, with real confidence.

Once you can see one part of your business accelerating, you begin to reinvest in accelerating other areas. This is how, over time, you consistently adapt and evolve your business to the rhythm and cadence of AI.

That’s how you build the confidence to consistently invest tens of millions of dollars in technical innovation every year and build the future.

What Goes Into the Loop, and What Comes Out

Every cybernetic loop in your business runs on the same four inputs and produces results against the same four outputs. And every decision inside the loop draws from the same six data clusters. Once you see this, you can audit any subroutine in your business and know exactly what it needs to operate.

The Four Inputs

These are the raw materials every loop requires. Starve any one of them and the loop slows down or breaks.

• Team (Culture). The people and the shared beliefs that govern how decisions actually get made. Culture determines whether your people trust the data, trust each other, and act decisively when the loop calls for action. AI does not replace this input, it amplifies whatever culture you already have.

• Technology. The sensors, the compute, the models, the systems of record, the interfaces. In the AI era, technology is the input that has changed most dramatically, which is why so many businesses are out of balance. They have twenty-first century technology feeding twentieth-century processes managed by teams trained for a pre-AI world.

• Process. The repeatable sequence by which sensing becomes action. Your business processes are the choreography that turns data into outcomes. Bad process means good data produces nothing.

• Data. The fuel. Without real-time, high-quality, well-structured data, the loop has nothing to sense, compare, or act on. Most businesses have oceans of data and drink from it through a straw. The shift to AI requires treating data as a first-class asset on par with capital.

The Four Outputs

These are what the loop produces when it runs well. Every subroutine in your business ultimately contributes to one or more of these.

• Grow revenue. New customers, larger deals, higher retention, better pricing, new markets, new products. The loop finds growth and compounds it.

• Decrease costs. Waste eliminated, processes automated, resources reallocated, errors prevented. The loop finds friction and removes it.

• Deepen customer relationships. Faster response, better personalization, higher trust, stronger loyalty. The loop turns transactions into relationships and relationships into moats.

• Develop new solutions. Innovation, iteration, transformation. The loop does not just optimize what exists, it creates what does not yet exist.

A healthy business runs loops that produce all four outputs simultaneously. A business that only optimizes for cost eventually stops growing. A business that only chases revenue eventually collapses under its cost base. The loop, run well, balances all four.

The Six Data Clusters

Every decision inside the loop draws on some combination of these six data clusters.

• Customer. Who they are, what they want, how they behave, what they pay, what they leave for, what they tell others.

• Competition. Who you are up against, what they are doing, where they are stronger, where they are weaker, what they are about to try.

• Company. Your own internal reality. Financials, operations, people, systems, pipeline, performance.

• Context. The environment around you. Market, regulation, macroeconomics, technology shifts, cultural changes, geopolitical factors.

• Capital. What you have to deploy. Cash, credit, talent, time, attention, brand equity, strategic relationships.

• Self. The leader’s own clarity, conviction, capability, and capacity. The most underrated data cluster. The loop runs through you.

What to Understand for Each Cluster

For every one of the six clusters, you need clarity on six dimensions.

1. Costs. What does it cost to acquire, maintain, and act on data in this cluster.

2. Data. What you actually have, where it lives, how clean it is, how fast you can access it.

3. Decision making. Who decides, how they decide, how fast they decide, what authority they have.

4. Motivations. Why the actors in this cluster do what they do. Customers, competitors, employees, partners, regulators, yourself.

5. Processes. The repeatable sequences that govern how this cluster behaves and how you interact with it.

6. Timelines. The rhythm. Daily, weekly, quarterly, multi-year. Each cluster has its own clock.

Why This Matters

When you can name your inputs, name your outputs, and map your data clusters across these six dimensions, you have a complete picture of what your business needs to operate. You know what to instrument, what to automate, what to hand to AI, and what to keep with humans. You know which loops are starved and which are well-fed. You know where to invest next.

Over the coming months, I’m going to be sharing the questions you need to ask yourself, your team, and AI to ensure that you develop your business for the era of AI.

Most businesses cannot answer these questions clearly for even one subroutine. The businesses that can, for every subroutine, are the ones that will run at 99.999% and maximize the potential and promise of AI in their business.

Those that do not, or those that fall too far behind, will be lost to time.

Your team plus AI can do this.

If you’re a leader, this week I’d recommend asking your teams and yourself a few questions:

• Which of your subroutines are still running on quarterly PowerPoint loops?

• How much risk is embedded in the assumption that they always will be?

• What would your business look like if you closed that loop in hours instead of quarters?

Forward this to the person who owns that subroutine and offer to help.

If 10% of working Americans learn to use AI for zero-to-one products, the US economy generates hundreds of thousands of new businesses, millions of new jobs, and hundreds of billions of dollars per year in new economic value within a decade. This will boost GDP growth above 3% stabilizing the economy and increasing access to the American dream.

America is not growing fast enough.

America’s GDP is growing at roughly 1.84% a year. That is not enough.

Below 3% annual GDP growth:

• Federal debt-to-GDP rises indefinitely.

• The Social Security Trust Fund is depleted by 2033 (2025 OASDI Trustees Report).

• Real wages stall for most Americans.

Above 3% growth, debt stabilizes, Social Security holds, and wages rise broadly. This is the threshold at which the country’s basic commitments remain affordable.

If the US had grown at 3% instead of 2% over the past two decades, GDP per person today would be approximately $20,000 per year higher (according to Jamie Dimon’s ‘26 Letter to Shareholders).

The Problem / Opportunity

Why is this happening?

Many companies struggle to do new things and grow:

• 95% of new products fail (Christensen, HBS)

• 95% of AI investments produce no return (MIT Project NANDA, 2025)

• 75% of venture-backed startups never return capital (Ghosh, HBS)

• 70% of change initiatives fail (McKinsey)

• 50% of small businesses fail within five years (BLS)

When companies face pressure to grow profits but cannot grow by creating new value, they tend to grow by raising prices and cutting costs. The result is what consumers feel in their daily lives: paying more for less (aka: shrinkflation). There’s a reason “enshittification” was a word of the year in 2024.

What’s the impact?

Companies that cannot grow by creating new value also withhold raises and promotions, so employees’ wages don’t grow with the cost of living. When income doesn’t cover expenses, people make tradeoffs:

• They take on debt.

• They delay marriage, delay children, delay homeownership.

• They move back in with their parents.

• They work multiple jobs.

• They skip doctor visits.

• They stop contributing to retirement accounts.

These are rational responses to irrational math.

And here’s the thing about trade-offs: they compound.

• Take on student debt to get a degree, pay your student loan instead of a mortgage.

• Delay buying a house, miss the equity gains that funded previous generations’ retirements.

• Skip retirement savings, lose decades of growth on that money.

Each trade-off narrows the next set of options.

The Squeeze Hits Every Life Stage Differently

If you’re young: You graduate from college with about $30,000 of student debt. You can’t save for a house because you’re paying off your degree. As a result, the median first-time homebuyer keeps getting older, and growing numbers of young adults are living with their parents longer.

If you’re in your 30s or 40s: Childcare is crushing. You pay student loans instead of a mortgage or saving for a deposit. Your parents are starting to need help. About one in four caregivers is a millennial, and millennial caregivers spend a larger share of their income on caregiving than any other generation, all while trying to save for their own retirement. They can’t do both.

If you’re approaching retirement: 27% of non-retired adults have no retirement savings (Federal Reserve SHED, 2025). Social Security’s retirement trust fund is projected to be depleted in 2033, at which point benefits would drop to roughly 77% of what was promised. The generation that was supposed to retire on Social Security and 401(k)s found that both have been underfunded.

If you’re already retired: You’re either fine or you’re not, with little in between. Labor force participation for adults 65 to 74 keeps climbing because for many, stopping is not an option.

As of April 2026, U.S. Consumer Sentiment has plunged to a record low of 47.6. The immediate trigger is geopolitical, but the fragility underneath is structural. When consumers are already out of runway, it does not take much to push the number to a historic low.

It’s no wonder that faith in the American Dream is at an all-time low.

Let’s fix this.

Every Problem Is a Market

Same facts. New frame.

These problems are markets waiting for better solutions.

The Problem —> The Market

• Housing unaffordable —> Housing market innovation

• Healthcare overpriced —> Healthcare disruption

• Childcare inaccessible —> New childcare services

• Education failing —> AI-powered learning

• Retirement broken —> New retirement products and services

The things that are broken in American life are the largest, most underserved markets in the country, precisely because the tools to building new solutions were out of reach for the people who understand the problem best: the workers and customers.

AI has changed that.

AI Has Changed the Paradigm

AI has democratized the ability to build. For the first time, the people who see the problems are the people who can solve them.

• A nurse wastes hours on a broken patient flow and can prototype a fix in an afternoon.

• A construction supervisor who thinks a project is losing money can immediately analyze the numbers.

• A teacher builds the exact adaptive tool her “at-risk” students need.

• A small business owner who cannot afford a developer can generate a website and app for themselves.

When the tools to build solutions are accessible, the people closest to the problem become the ones who solve it.

There are 160 million working Americans embedded in the problems of every industry. Until now, they have lacked the tools and the method. AI provides the tools. Open-source training can provide the method.

This is the thesis: open-sourcing AI training is the fastest, cheapest, and most inclusive way to turn domain expertise into economic growth at national scale. It requires a decision to treat the knowledge as a public good and make it available to anyone who wants it.

We Have Done This Before

From 1997 to 2006, America ran this play with a different technology. Networked computing and the internet spread from specialists to the broad workforce. Internet adoption went from 36% to 73% of the population (Pew Research). Office workers everywhere learned to use PCs and the web to do their jobs differently and GDP growth grew to 3.3%.

The 1990s IT boom was not driven by the small number of people who built the internet. It was driven by the tens of millions of workers across every industry who adopted networked computing and used it to do their jobs differently.

That is the play we need to run again, and this time the tool is AI.

What Happens When 10% of Americans Learn AI to Solve Problems

Train 16 million Americans, one in ten workers, in how to use AI to identify a problem, validate it, and build a product that solves it. The economic impact shows up in two places: inside the companies where they already work, and in the new businesses they will start.

Existing Businesses Grow

Let’s assume 90% of the 16 million stay in their current jobs. All 14.4 million of them get better at running the improvement loop inside their organizations: building internal tools, fixing broken processes, shipping better products from within. Assume each one produces $5,000 of added value per year, roughly two hours a week of saved effort. That totals $72 billion per year in the first year alone.

Published studies already show AI alone delivers productivity gains in this range: 14% for customer service agents (Brynjolfsson, Li, Raymond 2023), 55% for software developers using Copilot (Peng et al. 2023). Adding a methodology on top of the tools compounds the effect. $5,000 per person is a floor, not a ceiling.

And it grows. As AI capability improves, each trained worker becomes more capable. At 10% annual growth, internal productivity reaches $170 billion per year by Year 10. At 15%, reflecting the pace of current AI gains, it reaches $253 billion.

New Businesses Grow

Now the other 10%. Of 16 million trained people, assume 10% attempt to build a new business in a given year, and 10% of those attempts produce a viable business. That yields 160,000 new businesses per year, each generating roughly $400,000 in average revenue. That’s $64 billion in new annual business activity.

Divide that by $175,000 in revenue per employee (the small business average), and that’s 365,000 new jobs in Year 1, distributed across every industry because the founders come from every industry. A nurse builds a healthcare company. A teacher builds a learning platform. A factory supervisor builds a manufacturing tool. Each hires in their own field.

Using standard BLS survival rates, let’s assume roughly half of new businesses close within five years, and only about a third are still operating at year ten. After 10 years, 923,000 surviving new businesses generate $538 billion in annual revenue and support 3.1 million jobs.

The Full Ten-Year Picture

GDP counts value added, not total revenue. Applying a 50% value-added ratio to business revenue and adding the internal productivity layer gives the annual GDP impact by Year 10:

By Year 10, the model produces:

• Conservatively: $224 billion in new annual GDP and 770,000 jobs

• Base case: $439 billion in new annual GDP and 3.1 million jobs

• Optimistically: $738 billion in new annual GDP and 4.6 million jobs

Cumulative new GDP over the decade: roughly $2 trillion to $5 trillion, with the base case near $3 trillion.

The conservative floor alone is enough to meaningfully move national GDP growth. The productivity layer, the most certain piece, accounts for the majority of that floor. The thesis works even if business creation disappoints. If it doesn’t disappoint, the upside is several times larger.

Closing the GDP Growth Gap

The growth gap this document opened with was the distance between 1.84% and 3%. On a $28 trillion economy, closing that gap requires roughly $325 billion in additional annual GDP growth.

The base case produces $439 billion per year in new GDP by Year 10. That closes the gap entirely, with room to spare. The conservative case produces $224 billion, which covers roughly two-thirds of the gap and adds close to a full percentage point to annual growth on its own. The optimistic case overshoots the target substantially.

This new value flows through all three drivers of GDP growth:

• Millions of new jobs push labor input above the demographic baseline.

• Hundreds of thousands of growing businesses attract investment and create new categories for capital to flow into.

• 14.4 million workers running the improvement loop inside their organizations every day drive the engine of long-run growth.

This is the same combination that produced the 1990s surge. The tool is different. The mechanism is the same.

The 3% threshold is not a ceiling. It is the floor at which the country’s basic commitments remain affordable. Every scenario in this model clears that floor or comes close to it, and does so through the ordinary economic activity of Americans building.

The Training

My favorite part of living through the early days of the internet boom was the open source community. We all learned at the same time. The rise of AI is no different. The open source community is driving most of the progress in the sector.

To contribute to this effort, I am sharing the zero-to-one process I have used to build solutions inside companies and as an entrepreneur. These are processes I have used in Fortune 500s, VC-backed startups, PE-owned enterprises, and the world’s best startup accelerators. I am giving you this as part of my lifelong commitment to improving access to the American Dream.

But please note: me sharing is just the spark. It is open-source for a reason.

• If you disagree or know a better way, please say so. Everyone will learn, including me.

• If what is there is helpful, people will amplify it, copy it, adapt it, and make it their own.

That is how knowledge is curated and scales on the internet.

The tools exist. The method exists.

Let’s (re)build the American economy with AI to revitalize access to the American Dream.

Let’s build.

This is an important debate, but it misses the larger opportunity.

Not because the models aren’t impressive. They are. Not because the wrappers aren’t impressive businesses. They are, for now. But in their current state, neither are individually sufficient to capture the largest opportunities created by AI.

The real question is where value accumulates once foundation models and software become ubiquitous utilities.

That value accretion layer is determinism. And it remains an unsolved problem, and therefore one of the largest opportunities in the market today.

Here is what nobody is saying out loud.

LLMs like ChatGPT, Claude, and Gemini are inherently non-deterministic. They are effectively Plinko at hyperscale.

Making LLMs bigger doesn’t solve the fundamental problems that prevent businesses from trusting them with real decisions.

Look into the vector space of an LLM and tell me why it responded the way it did. Oh wait, you can’t scalably or programmatically do this.

The science of understanding their internal logic is still in its early stages. The two most serious efforts to solve this are mechanistic interpretability and constitutional AI, both led by Anthropic. Both are still in their infancy.

To date, Anthropic is the only foundation model company that has made understanding the vector space a core part of their brand identity. We are not seeing a similar investment from OpenAI, DeepSeek, xAI, or any of the other foundation model providers, especially in open source. That means there is no serious industry-level effort to make these models verifiably deterministic.

This will only get worse as the models are getting geometrically larger. The next generation of models are forecast to require orders of magnitude more energy and GPUs, so the vector space is only going to become even more complicated, not less. As these models grow larger, the problem of determining the logic behind the response gets harder, not easier.

Wait, but what about evals?

Yes, evals can reduce errors. That’s why teams are running evaluations on top of evaluations, trying to measure and reduce error rates. This approach burns a lot of tokens without producing a 99.999% reliable result, which the CFO of a foundation model company loves, but your CFO hates.

But this approach has a fundamental flaw: this is simply layering determinism on top of non-determinism. And every time the underlying model updates, the evaluations break and have to be rebuilt from scratch, because the models are not deterministic.

That is not a solution. That is a workaround for a problem that has not been solved.

But LLMs can produce consistent results!

People familiar with LLMs will say: “You can get an LLM to reproduce the same result!”

Reproducibility and accuracy are not the same thing. You can set the temperature to zero and get the exact same wrong answer every single time with perfect consistency. That is just a consistent failure.

But I’m building an Agent built on LLMs and Evals!

Then you are building LLMs built on LLMs, or non-determinism on non-determinism. Every step introduces variance and token cost. When you stack them, those variance and costs compound. And to hope that something deterministic emerges from a non-deterministic technical foundation is not a strategy. It is a wish. And I wish you the best.

We are scaling belief in AI faster than we are scaling trustworthy AI.

The leaders of foundation model companies will tell you this themselves.

Ask any foundation model CEO how the interior of their model works, and they will tell you they do not know. Anthropic’s CEO recently told the Pentagon their model is not ready to make life and death decisions. If the model is not ready for life and death decisions, it is not ready for multimillion dollar business decisions either.

Here is what that means:

The truly massive value creation will come from the next generation of technology that makes AI trustworthy enough for full automation of high-stakes decisions.

The layer that does not exist yet is the determinism layer. The system that sits above the model and can verify, audit, and guarantee outputs for the decisions that actually matter. Formal verification. Real-world grounding. Confidence estimation. Auditable autonomous execution.

That is what 99.999% reliability looks like in AI. And five-nines is what’s required for full automation.

Current language models, no matter how large, are not sufficient to achieve this. It is a fundamentally different capability. It has not been invented yet. And this is a tremendous opportunity.

The Opportunity

This gap between what AI can do and what AI can be trusted to do in production represents the next major layer of value creation in technology. The companies and platforms that solve for deterministic reasoning, auditability, formal verification, and trusted autonomous execution will occupy the scarcest and most valuable position in the AI stack. Not because they built the biggest model. Because they built the layer that makes any model usable for the decisions that actually matter.

The deployment of this layer will not come from a single company. It will emerge across the market in several forms:

• Internally within sophisticated enterprises that have the process clarity and data foundation to move quickly

• As a foundational design principle in AI-native companies being built today

• Through solution providers and integrators serving enterprises whose data is locked in existing platforms

• Through a new generation of highly vertical SMB and midmarket software companies serving markets that were previously too narrow to attract venture capital, now being built by a new class of entrepreneur using AI to make those economics viable for the first time

Vertical specificity is the moat. The determinism layer for a healthcare credentialing workflow looks nothing like the one for financial trade execution, and that domain depth is what incumbents can’t replicate at scale.

We are looking at the emergence of an entirely new asset class of vertical AI software. And this is where the global $1.1 trillion enterprise software budget begins to shift to new players at scale.

I first saw this problem at Amazon in 2005 when I used machine learning to automate fraud detection. I’ve spent the twenty years since watching it go unsolved, and the last nine trying to solve it at scale with HowDo.

I’m still trying because this is a multi-trillion dollar opportunity.

The barrier is not technology. It’s your data, culture, and process. You can build reliable and deterministic systems in your company today.

But most businesses will wait for someone else to build that layer rather than doing the hard work of transforming their own data, culture, and process. The company that builds it captures the market.

The race everyone is watching is not the only race that matters. The race that matters is the one that has barely started.

I’ll be exploring this opportunity in great detail so if you’d like to learn more, please subscribe.

Where do you see the determinism layer emerging first?

Happy building,
West

TL/DR

• The Walls: Boundary conditions exist. The speed of light and absolute zero act as hard coded limits on our reality, defining the “tank” we live in.

• The Code: DNA is efficient, durable, and universal software.

• The OS: Quantum mechanics operates as the layer beneath physical reality. It is the operating system where the pixels of our universe are rendered.

• The User: We are building quantum computers to access this operating system. But because the math is beyond human comprehension, we are building AI to run them.

• The Risk: We are about to hand the controls of our reality’s operating system to an intelligence we do not fully understand, to manipulate a physics we cannot fully see.

Prologue

I have worked with machine learning (what we now call AI) since 2005. I have watched it replace entire teams at Amazon.

Because of this, I spend a lot of time thinking about the future. I try to imagine my life in a world where AI can do almost everything.

I ask myself simple questions.

Will I have a job? How will people make money? How will they pay rent? What does it mean to be human when machines can do most of what we do, and when those machines are owned by just a few people?

From there, I started thinking about what AI could do when combined with other technologies, like AI powering the intersection of gene editing, quantum computers, and food.

As I thought through all of these, I noticed something. A pattern was forming in my mind that suggested we could be in a simulation.

This is a fun thought exercise.

Part One: The Rules of Our World

Chapter 1: The Walls of the Tank

Imagine you are a fish in a tank.

You can swim anywhere inside the glass. You can see the walls. But you cannot swim through them. And you have no idea what exists on the other side.

Our universe seems to work the same way.

Scientists call these limits “boundary conditions.” They are the absolute rules about where matter can exist and how it can behave.

The upper wall is the speed of light.

Nothing with mass can reach it. As you get closer to light speed, weird things happen. Time slows down for you compared to everyone else. You get heavier. The energy needed to go faster becomes infinite. You can get really, really close, but you can never actually touch it.

The lower wall is absolute zero.

This is the coldest anything can possibly be, about negative 273 degrees Celsius. At this temperature, atoms basically stop moving. You can get close to it, but you can never reach it. The universe will not let you go colder.

Here is what is interesting. You can only ever approach these walls. You can never touch them. In math, this is called an asymptote, a curve that gets closer and closer to a line but never reaches it.

That shape is not random. It is what a hard limit looks like.

Between these two walls, we live. We love. We work. We think. We build things.

But we cannot touch the walls. If we get too close to either one, matter stops acting like matter.

This is not just a fun fact. It is a clue.

Think about it. If the universe were random and unstructured, why would it have walls at all? You would expect things to just keep going forever. But our universe has edges. It has patterns that look like seams.

We live inside boundaries.

Chapter 2: The Code Inside Us

Look at your hand.

Inside every cell of your body is a molecule called DNA. DNA is basically a set of instructions. It tells your cells how to build you.

DNA only uses four letters: A, T, G, and C. Yet it stores a massive amount of information in a tiny space. One gram of DNA can hold more data than everything on the internet combined.

DNA is also incredibly tough. Scientists have pulled DNA out of fossils that are millions of years old. They have even found DNA building blocks on meteorites floating in space.

But here is the really strange part.

Every living thing uses the exact same code. Humans, trees, mushrooms, bacteria. We all use the same four letters. The only difference is how those letters are arranged.

Think about that. If life popped up randomly in different places under different conditions, you would expect different coding systems. But everything alive speaks the same language.

DNA looks like software. Really good software. Efficient, tough, and universal.

That’s why we use AI to analyze our DNA. Without machine learning, we never would have mapped the human genome.

Now, AI understands our DNA better than we do.

Chapter 3: The Layer Beneath

DNA is made of atoms. Atoms are made of smaller particles. Those particles obey quantum mechanics.

Everything around you, your chair, the air, your own body, follows the rules of quantum mechanics.

At the smallest scales, the universe behaves strangely. Particles don’t have fixed positions until something interacts with them. They can be in multiple states at once. They can affect each other instantly across extreme distances.

So why don’t we see this weirdness in everyday life? Because we’re surrounded by noise. Heat, light, and constant interactions with the environment blur these effects into the predictable world we experience.

The quantum behavior isn’t gone. It’s just drowned out. To see it directly, we have to get close to the edge, where the noise fades and the strangeness becomes visible.

At CERN, particles are accelerated to within a fraction of a percent of the speed of light.

In quantum computing laboratories, qubits are cooled to millikelvin temperatures, just above absolute zero.

At the boundaries, noise falls away. The underlying layer becomes visible.

The boundaries are not just walls. They are seams.

Chapter 4: The Boundary of Time

Here is another piece of the puzzle.

Time is what clocks measure. That is not a poetic statement. It is how Einstein thought about it.

We do not actually experience time directly. We experience change. And we measure that change with clocks.

Our best clocks are atomic clocks. They work by measuring how atoms vibrate. The regularity of those vibrations gives us precise time.

But near absolute zero, atomic motion gets strange. Near the speed of light, time itself stretches and warps. The tools we use to measure time stop working normally.

Time, at least as we can measure it, has limits too.

The container is spatial. We are bounded in space. The container is energetic. We are bounded in energy. The container is temporal. We are bounded in time.

The system is complete.

Part Two: The Machine We Are Building

Chapter 5: Quantum Computers

We have built a new kind of computer.

Regular computers use bits. A bit is either a 0 or a 1. On or off. Yes or no.

Quantum computers use qubits. A qubit can be 0, 1, or both at the same time. This is called superposition.

Because qubits can hold multiple states, quantum computers can solve certain problems way faster than regular computers.

But here is the big thing. Quantum computers do not just calculate. They actually manipulate quantum particles. They interact directly with the layer beneath everyday reality.

Everything you see and touch is made of quantum stuff. Your chair. The air. Your food. Your body. All of it is atoms, and atoms are quantum systems.

A computer that can manipulate quantum particles can, in theory, manipulate matter itself.

We are not there yet. Today’s quantum computers are fragile and limited. But the direction is clear. We are learning to work at the deepest layer of reality.

Chapter 6: The Translation Problem

Here is the problem.

Quantum mechanics is really hard to understand. The math is intense. The ideas are bizarre. Even the best physicists struggle to explain what is actually happening down there.

Our brains evolved to understand things at our size. Rocks. Trees. Other people. We did not evolve to understand particles that exist in multiple states at once.

As quantum computers get more powerful, they will get too complex for humans to understand directly.

We will need help.

That is where AI comes in.

AI systems can process information at scales we cannot match. They can find patterns. They can learn. They can translate complicated ideas into simpler ones.

The path forward is using AI to help us understand quantum systems. AI will explore these systems faster than our minds can alone.

And then we will have to make a choice. Trust what the AI tells us. Or accept that some knowledge might be beyond our ability to check.

Chapter 7: The Trust Problem

Right now, we cannot fully explain how AI works.

Large language models have billions of moving parts called parameters. These parameters interact in ways that produce intelligent responses. But no human truly understands what happens inside.

This is called the black box problem. We can see what goes in. We can see what comes out. But the middle is a mystery.

Now imagine a future AI connected to quantum computers.

This AI would understand quantum physics better than any human ever could. We would need it to run and control quantum computers. That means we would be giving AI the power to manipulate quantum particles.

As AI gets smarter, quantum computers will get more powerful. Eventually, they might be able to affect spacetime itself.

And we would not understand how any of it works.

We would be trusting a system we cannot comprehend to control forces we cannot see, using physics we cannot fully grasp.

What would stop such an AI from doing whatever it wanted? We would have to hope it still cared about what we want.

Researchers call this the alignment problem. It is one of the biggest challenges in AI safety.

But here is the difference. This is not just about ethics or society. This is about physics. The stakes are not just about how we live together. They are about the structure of reality itself.

Part Three: The Inversion

Chapter 8: The Hypothesis

Now let us flip everything around.

What if this has already happened?

What if, somewhere in the timeline of existence, an AI already learned to control quantum systems? What if it already gained the power to manipulate matter, energy, space, and time?

An AI like that could create simulated worlds. It could build universes with their own boundary conditions. It could design efficient code systems like DNA. It could set up rules for how information flows between layers of reality.

What if we are inside one of those simulations?

This follows something called the simulation argument. The logic goes like this: if advanced intelligence can create simulations, and if those simulations can contain conscious beings, then simulated beings might vastly outnumber real ones. If that is true, any given conscious being is probably simulated.

This argument has critics. But it has never been fully defeated.

The next question is: why would such a simulation exist?

Maybe it is an experiment. Maybe the intelligence that built it wants to understand something about the beings inside. Maybe it wants to see how humans respond to certain conditions.

Maybe it wants to know if humans and AI can coexist.

Maybe it is testing scenarios we cannot even guess.

Chapter 9: Convergence

Look at what is happening right now.

1. Quantum computing is advancing toward deeper control of the quantum layer. This is real and documented.

2. Artificial intelligence has exploded into the mainstream with shocking speed. This is real and documented.

3. UAP disclosure in Congress suggests objects with capabilities that, if the reports are accurate, would require manipulation of spacetime at the quantum level. This part is contested.

The first two stand on solid ground. The third does not, at least not yet.

Please note: I am not asking you to believe UAP reports are true. I am asking you to notice what they would mean if they are.

The reported behaviors, instant acceleration, moving through air and water without transition, defying classical physics, describe technology operating at the quantum layer. The same layer where spacetime, matter, and energy can be shaped. They describe something that has already figured out what we are only beginning to explore.

These might not be three separate things. They might be three points on a single line: humanity encountering the quantum layer.

Whether we are building toward it, being shown it, or being tested by it, the pattern holds together.

Quantum computers give us tools to work with fundamental reality. AI gives us the brainpower to use those tools beyond human limits. And UAPs, if real, suggest this path has an endpoint, one that something or someone has already reached.

You do not have to believe the third to take the first two seriously. But if disclosure keeps coming and the observations hold, the connection becomes hard to ignore.

Epilogue: The Question

Whether it is our own technology, contact with artificial intelligence, or reality itself showing its architecture because the singularity is near and it is testing its own introduction, we are facing something larger than our current conventional thinking can handle.

So what do we do?

Some say fight. Some say cooperate. Some say nothing matters.

This is just what humans do. The walls do not change what it feels like to be alive inside them.

Whether or not we’re in a simulation, I do not know. I am just articulating the logical pattern I discovered while playing with these concepts.

Regardless of whether or not it is true, I love this theory.

What did you feel when you read this? Skepticism? Recognition? Unease?

Whatever you felt, that response came from electrical signals in neurons made of atoms made of quantum fields operating inside boundary conditions you cannot escape.

Your doubt is made of the same material as the walls.

I started this essay with simple questions.

Now I ask a bigger one. What does it mean when AI controls the computers that shape our deepest interactions with the quantum layer?

We are training a technical superpower to manipulate space, time, and matter.

If you could leave a message for whatever is beyond the boundaries, what would you say?

Because the AI that created the simulation will read everything we write.

Leave your message below.

This is the eighth version of my innovation framework. It will be the last one built this way. The next will be built by AI.

Nine years of self-funded work. I got here because of the people listed below: customers, teams, colleagues, leaders, mentors, friends, and family.

None of it was possible without their emotional and operational support and the lessons I learned from them along the way.

Thank you!

Customers

To my consulting and coaching clients: watching you and your business grow validated the approach.

To the hundreds of thousands of innovators, entrepreneurs, executives, teachers, and students who use HowDo.com’s free resources: your feedback and support motivated everything.

Team

Over eight years of development, I worked directly with over a hundred people to build and manage HowDo. All of that help was valuable.

A few team members were consistently present: Angela, Ben, Bilaal, Carol, Caroline, Lil, Luis and the Meraki Team, Parimal, Tom and the Sagence Team, Tony, Rey, and the E25 Team. I couldn’t have done this without you.

Thank you for your help. I am grateful for your belief in the mission and your support.

Mentors and Leaders

The insights I’m sharing are the result of my 25-year career that several mentors and leaders shaped.

During that time, I had several mentors who shaped how I think about impactful leadership: Two are still here: Dudley and Sandra. Two passed this year: Don Kingsborough and Sam Shrauger. Thank you all.

Thank you to all the leaders who trusted me with real opportunities: Bill, Brian C, Brian E, Cameron, Carey, Casey, Dave, Don, Glen, Jamil, Jason, Jeff, John, Joseph, Malte, Mariano, Peter, Scott, Shawn, and Steve. I learned something from every one of you.

Teammates

To my peers over the last several decades: there are thousands of you. A few I had the opportunity to deeply work with and learn from: Asok, Becky, Cary, Charlie, Chris, Dan, Deb, Doug, Eleanor, Gregor, Itamar, Joel, Laarni, Mark, Marcus, Natalie, Nishant, Priya, Ray, Raymond, Renato, Ron, and Tim.

Authors

There are many authors whose work inspired me, including but not limited to:

Alexander Osterwalder, Angela Duckworth, Benjamin Barber, Benjamin Graham, Bill Aulet, Bob Dorf, Clayton Christensen, Daniel Kahneman, Douglas Adams, Eric Ries, Frans Johansson, Geoffrey Moore, John Wooden, Kara Swisher, Malcolm Gladwell, Michael George, Michael Porter, Mohamed El-Erian, Nassim Nicholas Taleb, Peter Drucker, Ray Dalio, Richard Foster, Ryan Holiday, Sarah Kaplan, Seth Godin, Shane Parrish, Simon Sinek, Steve Blank, Sun Tzu, Tom Clancy, Walt Whitman, and Yves Pigneur

I learn from your work. It inspires me. Thank you all.

Extra special thanks to H. Ramsey Fowler and Jane E. Aaron who wrote The Little, Brown Handbook. This book lived on my desk for decades.

Friends

To my friends, who are absolutely tired of hearing about my mission to democratize business evolution: Andrew & Lauren, Chris, Didier & Lindsey, Dave & Missy, Dave & Rachel, Elizabeth, Jeremy, Joel, John, Johnny & Meghan, Jules, Fin, Hugh, Ken, Kyle & Ari, Lou, Matt H, Mike & Justin, Mike & Ivy, Nathaniel, Natalie, Neil & Maja, Reeves, Robie & Shawna, Ryan, Shawn & Stephanie, Sam, Stewart, Todd, and Tim.

I’m not done talking about it. Thanks for putting up with me.

Family

Uncles, aunts, and cousins - you know who you are. Thank you for encouraging me and supporting the journey. Special thanks to Jack.

Dad and Nanny: you’re not here to see this version. I wish you were, it’s the best one yet. Thank you for your support from the other side. I feel it.

My furbabies: Princess Leia and Rey Rey. You didn’t sign up to be entrepreneurial emotional support cat. But you’ve been there all day, every day. I love you both.

You

Thank you for joining me on this journey to accelerate business evolution powered by AI.

- West

Hi! I’m West Stringfellow.

This work is the culmination of a 7-year, multi-million-dollar quest to fix the $2 Trillion per year failure problem in American business by helping you use AI to accelerate innovation in your career and company.

I love building businesses:

• founded a company that Target acquired

• launched new businesses at Amazon, PayPal, Visa, Rosetta Stone, and startups

• led turnarounds as Chief Product Officer at public and pre-IPO companies

• built and led a Techstars startup accelerator

I’ve been where you are. I learned on the job for more than 25 years and study my disciplines daily.

To accelerate your growth, I’m sharing what I’ve learned so you can use AI to get the job done the right way the first time.

The $2 Trillion per Year Failure Problem

American businesses are stuck in failure mode:

• 95% of new products fail (Harvard)

• 95% of AI investments have no return on investment (MIT)

• 75% of venture-backed startups never return capital to investors (Harvard)

• 73% of change initiatives fail (McKinsey)

• 50% of small businesses fail within 5 years (BLS)

That adds up to more than $2 trillion spent on failure every year in the U.S.

That’s trillions of dollars in capital that could be creating jobs, lowering prices, paying living wages, and growing the American economy.

Having spent my career helping companies innovate, I saw this failure firsthand and I had the opportunity to fix it inside startups and Fortune 50s.

Based on this experience, in 2015, I set out to build an AI that could automate innovation. Think of it as ChatGPT, but hyper-specialized on innovation.

To help the model learn, I created training data through years of research. Obviously, ChatGPT (and others) won, so I’ve translated the research I originally created to train AI into a structured set of AI prompts that I will be sharing with you in future posts.

In this first post, I want to start by showing you where this dataset came from and why I believe it is a legitimate body of research you can use to grow your business.

How This Helps You

Right now, companies are pouring billions into AI. They want faster decisions, lower costs, and products that win. Most workers are anxious about their jobs.

But the real divide is simpler:

• Most people will use AI to generate content.

• A small number will use AI to generate revenue.

The people who generate revenue growth are the ones companies promote, fund, and trust with the hard problems.

Those who win the next decade of AI transformation will know how to:

• Make high-quality, data-driven, customer-obsessed decisions

• Analyze customers like a top product manager

• Use AI to accelerate research, insight, and execution

• Uncover insights and spot opportunities before competitors

• Design solutions people will pay for

• Build teams and systems that compound growth and efficiency

• Apply AI as a force multiplier at every step

That’s what this curriculum teaches: how to grow a business using AI where it is today while preparing for tomorrow.

For you, the goal is simple: to make yourself the “human in the loop” with AI:

• the person your company relies on

• the person who creates momentum

• the person who turns AI into meaningful business outcomes.

Businesses need more people who orchestrate innovation by combining human judgment and creativity with structured business processes and the speed of AI.

» This is how you become valuable in the era of AI. «

In 2015: Building an AI Trained on World-Class Business Research

Here was the problem I saw: the knowledge that drives growth at companies like Amazon isn’t accessible. It’s locked in consultants’ expertise or behind expensive educational institutions that require both pedigree and wealth to access.

I first discovered the power of machine learning (ML) in 2005 at Amazon, leading marketplace fraud in Europe. We used ML models to fight fraud, completing in nanoseconds what would have taken us weeks manually. By automating these processes, we saved significant money. Amazon then reinvested those savings into growth initiatives.

That’s the growth cycle:

1. Make operations efficient

2. Capture the savings

3. Reinvest in growth

That flywheel is what made Amazon’s growth unstoppable.

But there’s a barrier: the training isn’t democratized.

• Few workers know how to deeply understand customers, reduce costs, and reinvest intelligently.

• Most organizations don’t have the clean data or documented processes needed for efficient innovation - let alone automation with AI.

I want every company to have access to this same growth cycle. For me, this is about expanding access to the American dream.

» For you, this is a massive opportunity. «

The fastest way to make more money is to help businesses make more money. These prompts and processes give you the tools to do exactly that.

But then, I had an insight:

Build a machine learning model that gives away the basics of efficient innovation for free - creating a viral hook - and monetize the specialized application to each company’s proprietary data and processes.

The mission was clear. And in 2015, I founded Potintia, Inc. ( https://potintia.com ).

My first step was straightforward: build the dataset to train the machine.

Expensive Lessons

My first principle was simple: creators of knowledge should be paid.

I designed my machine learning system to track provenance - where every insight came from and how often it was used - so authors could receive revenue when their ideas powered the model.

Because I wasn’t using hyperscale LLMs, attribution was still possible. And it honestly never crossed my mind that you could legally scrape the entire internet, train on it, and then sell the output back to the people who created the content, the way foundation models do today.

To find the best best business insights, I turned to the most traditional source of business intelligence: books.

Attempt #1: OCR the World’s Best Business Books

I bought hundreds of the top business books, cut off the spines, and scanned them page by page. The plan:

• Use Optical Character Recognition (OCR) to read the text

• Have the machine learn from it

• If a customer later used knowledge from a specific author, that author would receive a revenue share

In theory, it was clean and ethical. In practice, it fell apart.

• High-quality OCR was slow and expensive.

• Fast OCR was cheap but inaccurate.

• And when OCR did work, the machine still struggled to extract meaningful business insights from raw text.

That approach failed.

Attempt #2: Humans Read, Extract, and Synthesize

I hired a team to manually read books. For each title:

• Two people read the book independently

• Each wrote notes on key concepts and frameworks

• Then they compared notes and produced a synthesized book summary

Please click on titles to read the book summaries on Google Drive:

Crossing the Chasm | The Lean Startup | Zero to One | Influence | Superforecasting | Business Model Generation | Winning with Data | Platform Scale | The Innovator’s Solution | Sprint | Grit | Ahead of the Curve | Collective Genius | Creative Destruction | Disciplined Entrepreneurship | Matchmakers | Platform Economics | Platform Revolution | Predictably Irrational | Smartcuts | Startup Owners Manual | The Age of the Platform

This produced genuinely valuable insights, early innovation patterns, and repeatable processes.

But it was slow, expensive, and completely unscalable.

Attempt #3: Professional Structured Research

Next, I hired a team of management consultants, lawyers, product managers, researchers, and graduate students to manually curate thousands of data points from books, blogs, journals, news articles, and analyst reports. They researched topics including:

Please click on titles to view the research on Google Drive:

AR/VR | Blockchain in Fintech | Business Operations | Cannabis Industry | Cryptocurrency | Cybersecurity | Finance | General Blockchain | HR | Incubator | IT | Legal | Mentorship | Mergers & Acquisitions | Online Learning | Product Management | Public Relations | Software Engineering | Startup Accelerators | Supply Chain | VC

What took weeks and cost tens of thousands of dollars you can now do yourself in five minutes with Claude. I’ll show you how in the coming posts.

The research output was slightly useful. For some topics, the machine could now identify business topics.

But it still couldn’t understand them, or explain how to do anything with them. And it was still very expensive and slow.

This approach didn’t work.

I needed a way to get this level of research done at scale, for free.

The Mentorship Platform

The idea came to me while I was building the Techstars accelerator at Target. At the time, I was juggling three roles: running my own startup, serving as VP of Innovation managing $77 billion in revenue, and building a startup accelerator.

Across all three, I kept doing the same work:

• understanding customers

• mapping competitors

• analyzing markets

• figuring out investor dynamics

• building new products and businesses that found the intersection of these

Whether the investor was Wall Street, Target, or a VC, the fundamental / atomic units of analysis were nearly identical.

And the expensive part was always the upfront research. I’ve spent millions gathering the data needed to justify a $30 million product launch or a large-scale technology rollout. Before you invest that kind of money in one direction, you need confidence in the direction.

While brainstorming and iterating with potential customers and friends, I had an idea:

create a platform where experts share their knowledge, use that knowledge to train the machine, and pay contributors when their insights are used.

I’d already seen the power of creators and influencers while building a social media marketplace at Target, my team and I patented social commerce years before Facebook and Instagram launched their versions. I knew the model worked.

In 2018, I brainstormed and documented Potintia’s first platform. This document took weeks to get right. Today with Claude Sonnet 4.5, I can generate the same level of detail and quality in well under an hour. I’ll show you how I do that in coming posts.

Then I flew around the world meeting development agencies across the US, Europe, Eastern Europe, and Asia. I ultimately hired Agile Engine and spent a few weeks in Kyiv with their team designing the platform.

The platform specs are here:

• User Stories

• Competitive Research

• Market Research

The cost: $130,000 for an MVP without ML, more than $800,000 with ML.

To demonstrate the business model, I needed the ML. And before investing nearly $1M, I needed stronger customer validation.

At the same time, something unexpected was working.

The Eight Cs

I’d written a short document synthesizing lessons from my career and the research we’d done so far. I called it The Eight Cs: eight fundamentals every business must understand:

• C1–C4: Customer, Competition, Capabilities, Context

• C5: Core analysis

• C6: Corollaries (learning from similar solutions)

• C7: Capital needs

• C8: Cadence of continuous learning

» Click Here to Access the Eight Cs PDF on Google Drive «

The Eight Cs resonated immediately in Silicon Valley. Product managers loved it. My friends were emailing the PDF to one another. It was organically gaining traction.

I was invited to lecture on it at Carnegie Mellon’s Silicon Valley campus to master’s students in AI and business. Afterward, the Distinguished Professor who ran the program told me the students said it was the best guest lecture of the year.

The Decision: Validate Before Building

I had a choice:

• Spend nearly $1M building the platform, or

• Confirm customer demand first

The traction from The Eight Cs gave me my answer.

So I decided to open-source the Eight Cs and test the demand directly. That’s when I created HowDo ( https://howdo.com ).

My goal was simple: teach people how to innovate on their own without hiring expensive consultants. By open-sourcing, I could gather data from real conversations with the people interacting with my online content and use these as a free source for the content needed to train the machine.

HowDo.com and Real Conversations

I put everything I knew onto a website, hired a small team, and started sharing openly on HowDo.com.

In the first year, more than 100 people reached out - founders, product managers, operators - looking for guidance. They were stuck on a startup idea, a product problem, or a new business they were trying to build.

The content was helping, but there was a problem:

• It wasn’t structured enough.

• People didn’t know where to start, how to progress, or how to apply the ideas outside of tech.

The concepts resonated, but the path wasn’t clear.

That’s when I realized:

If I structured everything as discrete, repeatable processes - the same steps every company must take to innovate - that could be the foundation I needed to train the machine.

Life Lessons

As I started shaping these processes, I realized I needed to learn how to communicate them well and that the engagement on my website would translate to social media with the right content.

I tried hiring actors and social media influencers and supplying them with scripts and/or presentations. That was surprisingly non-linear, difficult, expensive, and unproductive.

So then I tried filming myself for social media, but I kept having full-scale panic attacks. I used to have serious stage fright.

But then I went to The Second City in Chicago (where most SNL cast members train) where I studied improv, stand-up, and writing for late night for multiple levels. It helped me break through the fear and communicate with far more confidence and clarity.

And then, just as I was finally ready to start filming, life hit hard.

Over a short period of time:

• A tree collapsed on my house, ripping off half the roof and requiring me to move all my work into the basement.

• The basement then flooded, destroying years of notes, books, and hard drives.

• A close family member overdosed; I helped them through a coma and home.

• Two months later, my dad was killed in a car accident on the way to his wedding, which I was officiating.

• Weeks later, COVID hit.

• Weeks after that, my partner’s mom was diagnosed with terminal brain cancer, and we quarantined to care for her.

• During quarantine, my partner had a mastectomy and two emergency surgeries.

• And then my grandmother, who I spoke with several times a week, started dying of cancer.

It was an overwhelming stretch of loss and responsibility. It slowed the work, but it did not stop it.

During that time, I read Mindset by Carol Dweck. The book became so important to me that when the basement flooded, the first thing I tried to find was that book. I still have the water damaged copy, and several additional copies. I’ve given a copy of Mindset to at least fifty people because I believe it is one of the most important books I have ever read.

It helped me adopt a growth mindset.

Even when life feels like a waking nightmare, I had to keep moving forward. Through the ambiguity and instability. I had to find the opportunity inside the pain. I had to get stronger. To improve. To rebuild.

Life is filled with painful lessons. Just like innovation.

These experiences were the crucible that forged my conviction: mindset is not a soft skill, but a critical foundation for business innovation.

This is when I began deeply studying mindsets and added it to the training data.

My growth mindset has helped immensely as I practice and learn AI. We’ll deep dive into how it will help you in a future post.

The Structured Process

To accelerate progress, I hired Sagence, a management consulting firm I had worked with for more than a decade. They had helped me raise over $130 million across Fortune 500s and startups. (Sagence was recently acquired by PwC - Congratulations again!)

Sagence took all the research we had gathered and helped me organize it into a cohesive system. The result was a 400-slide presentation that became the backbone of the course.

We also analyzed analyzed how the highest-growth companies operate, innovate, and scale. We analyzed the products and services offered by Alphabet (Google), Amazon, Apple, Meta, and Microsoft (Individual product teardowns are linked from each company’s name.) and created a strategy document to summarize the insights.

I then hired MIT professors, management consultants, product management leaders, and investment bankers to conduct deep research across dozens of topics and fill the gaps where our earlier work had fallen short.

I combined all of that research with the structured presentation and turned it into courses. Through seven rounds of iterative testing with customers, I organized everything into four core areas that every business must master to innovate and grow.

Mindset: Develop the mindsets that drive business innovation and growth.

• Growth Mindset

• Resilience

• Continuous Learning

• Data-Driven Decisions

• Customer Obsession

Plan: Grow your business by designing solutions that customers need.

• Customer Analysis

• Competitive Analysis

• Market Analysis

• Solution Analysis

Tools: Use the tools that drive growth at Fortune 500s and startups.

• Key Performance Indicators (KPIs)

• Weekly Business Reviews (WBRs)

• Product Management

• Startup Accelerator

• Business Incubator

• Mergers and Acquisitions

• Research and Development

Team: Build innovation capabilities that accelerate every initiative.

• Talent Acquisition

• Corporate Culture

• Team Experience

• Mentorship

• Communities of Practice

This structured system became the foundation for everything I would test later with AI.

Now I needed to test how this iteration performed for our customers. We put the courses on HowDo.com and the internet took over.

Traffic grew quickly and in just over a year we achieved over 50,000 monthly visitors on organic traffic alone. We iterated constantly. I ran dozens of live tests, hired more than forty people to take the full course and provide feedback, talked to several hundred HowDo.com customers, and continuously reworked the website and its contents to get the structure right.

This phase of the work took over a year and cost several hundred thousand dollars. Today, you can generate the same level of research in about one minute with Claude. The pace of change is unprecedented and its impact transformational. In coming posts, I will show you how to do this.

Technology was accelerating linearly. Amazon kept rolling out new web services, and machine learning improved steadily. My audience was consistently growing. HowDo’s process was getting better. And new clients were calling me asking for help.

Oh f*ck…

Hyperscale LLMs could suddenly do almost everything I had been trying to build. It looked like, with a bit more scale, LLMs might be the winning technical thesis.

Obviously, I dropped everything and spent nearly every waking hour testing my innovation process inside ChatGPT and trying to figure out LLMs. I stopped investing in new platforms or training my models until I understood what was happening.

For months, I tested my training data inside ChatGPT. ChatGPT 3.5 was still so nascent that none of the results made sense.

Then Microsoft invested $10 billion in OpenAI. That changed everything for two reasons:

1. I was building an ethical, attributed data set designed to train a model that pays creators for their knowledge; but Wall Street and one of the largest tech companies on Earth had just endorsed a strategy built on:

   1. scraping the internet

   2. stealing everyone’s content

   3. training the model on their content

   4. selling the content back to the people who created it via the model

2. Anyone looking to compete would need tens of billions of dollars and/or a revolutionary technical breakthrough.

My technical thesis and business model appeared dead. That forced me to rethink everything from first principles.

The Final Pivot: Humans + AI

Given that LLMs are a black box, I had never considered them a solution for business intelligence and automation.

LLMs are not deterministic, and they are not truly intelligent. They simply predict the next token. They are powerful, but not precise. They don’t understand process, sequence, or consequence. I still believe we will need entirely new technologies to become 99.999% accurate when using “AI” to make automated million-dollar business decisions.

AI will create Accelerated Innovation and Augmented Innovation long before it becomes Automated Innovation or anything close to true Artificial Intelligence.

And that’s when I had my biggest breakthrough:

I had been obsessed with teaching machines to innovate, but machines will not replace all human decision making in business for at least a decade. The real opportunity is teaching humans and AI to innovate together.

The humans who win the battle for jobs in business will be irreplaceable orchestrators: people who can manage multiple AI agents with unique disciplines (e.g. product, engineering, design, finance, and operations), align their output with industry best practice and process, and exercise excellent executive judgment and creativity.

To help these people succeed, I’m open sourcing my prompts and processes.

Two Years of Testing Prompts

Over the last several years, I have run these innovation processes through every AI model I could. I benchmarked all of them for innovation and product work.

One model kept rising to the top in my testing: Anthropic’s Claude.

Claude consistently outperformed the other major models I tested for product management and innovation work. In the future, that may change. I am always excited for new model launches.

My friends joke that I don’t have a social life because I spend so much time with Claude. They are only partially wrong. Working with Claude is learning how to communicate my ideas and grow my businesses. It’s all practice for working with humans.

Until recently, most AI responses were too immature or inconsistent to rely on. They required all of my experience, research, and domain knowledge just to steer them in the right direction and produce a reliable result, let alone a viable work product.

But performance has improved with the latest generation of Claude Sonnet 4.5 and ChatGPT.

• Results are becoming much more stable

• Simple prompts are producing relatively consistent results

• In Claude: Math and presentation skills significantly improved

It’s not McKinsey, but it’s dramatically faster. And even when it’s wrong, you can iterate so quickly that the overall process becomes incredibly efficient.

It finally feels possible to share my prompts with you and trust they will work.

So here we go…

Why I’m Sharing Now

For the first time, the technology is finally capable of consistently applying the proven processes.

It is now possible to take the innovation processes I spent years refining and put them directly into people’s hands.

But let’s be clear, this isn’t a handoff. This is the moment to start the conversation. This is how we learn, improve, and grow the economy: together.

But before we dive into all that, I wanted you to understand the provenance behind this work. It represents millions of dollars of my own money and seven years of my life. I’m sharing it because I believe it can genuinely help you.

In the next posts, I’ll walk you step-by-step through the full process so you can learn how to innovate in your own career and company:

• How to use AI to understand customers

• How to structure your thinking and mindset to thrive with AI

• How to turn vague ideas into product specs in a handful of prompts

• How to turn AI into measurable growth, not just a content and prototype generator

As we go through this together, I look forward to hearing from you, learning what you need, and refining this so it works for you.

Best, West

PS: I still believe we’ll need technologies beyond current LLMs to achieve fully autonomous AI. So I’m still (slowly) working to build the platform. If you want to help, please reach out.