I was 24 when I picked up Bill Gates' Business @ the Speed of Thought at a Barnes & Noble.
I could not afford the hardcover, so I sat on the floor and read as much as I could before anyone asked me to move. Gates was writing about email, dashboards, digital records, and what he called a digital nervous system for business. The idea was simple and powerful: information should move through a company fast enough for the company to respond intelligently.
At the time, that felt ambitious. Looking back, it also feels like the first chapter.
Gates was describing a world where business could move faster because information moved faster. What AI changes is different. It does not just move information. It helps us work with it, question it, connect it, and turn it into decisions, plans, and artifacts.
That is a bigger shift. It is also a more dangerous one if we confuse speed with wisdom.
Gates was right about the bottleneck
The old bottleneck was access.
People made slow decisions because the right information was buried in email, locked in systems, delayed in reporting cycles, or scattered across departments. If you could connect the systems, surface the data, and shorten the feedback loop, you could make the organization smarter.
That thesis aged well.
Most companies still struggle with the same basic problem. The facts are somewhere. The work is somewhere. The context is somewhere else. People spend a surprising amount of their day trying to reconstruct what everyone already knows.
AI does not remove that problem by itself. In many teams, it makes the problem more visible.
Now every person has their own AI tab, their own prompt history, their own private working context. The work is faster, but not necessarily more shared. The organization may have more output and still have less alignment.
That is the part I keep coming back to.
The real shift is from information flow to thinking flow
When AI is useful, it does not feel like a faster search box. It feels like a thinking partner that helps you hold more of the problem at once.
It can compare notes, pressure-test assumptions, turn a messy transcript into a decision brief, summarize a tradeoff, or help you see where a plan is pretending to be clearer than it is.
That matters because the second bottleneck was never only information. It was cognitive load.
Even strong operators can only track so many threads. Founders can only hold so many moving pieces in their head. Teams can only re-explain the same project so many times before the signal gets thinner.
The promise of AI at work is not that every answer arrives instantly. The promise is that teams can keep more context alive, make fewer people start from zero, and convert more messy thinking into useful shared artifacts.
That is not "everything at the speed of thought." It is something more practical: fewer resets, fewer dropped threads, and better shared judgment.
Speed is useful. Speed is not the point.
I learned this the hard way as a founder.
At AEFIS, we worked on problems where speed helped, but speed was not enough. Accreditation, assessment, outcomes, institutional evidence, faculty workflows, governance. These were not problems you could brute-force with a clever dashboard.
You had to understand the work. You had to understand the people doing the work. You had to respect the fact that change inside institutions is not just a technical rollout. It is trust, language, incentives, and timing.
That experience shaped how I think about AI now.
Yes, AI can help a founder explore a market faster. It can help an operator turn scattered notes into a plan. It can help a team prepare a sharper meeting brief in minutes instead of hours.
But the best decisions still require judgment. Sometimes you need to wait. Sometimes you need to ask one more person. Sometimes the spreadsheet says one thing and the organization is telling you another.
The strongest use of AI is not replacing that human judgment. It is giving judgment a better surface to work on.
More ideas can become a trap
AI makes ideation cheap. That sounds great until you are staring at fifty plausible directions and none of them have been tested by reality.
Founders do not fail because they lack ideas. They fail because they pursue too many almost-good ones, or because they mistake a clean narrative for evidence.
AI can accelerate that mistake.
A model can generate a market map, a launch plan, a positioning statement, a feature list, and a fundraising narrative before lunch. Some of it may be useful. Some of it may sound useful. That distinction matters.
This is why I care so much about shared context and durable outputs. The value is not more private chats. The value is a workspace where the team can see the reasoning, challenge the assumptions, improve the artifact, and decide what happens next.
If the work cannot survive outside the prompt window, it is not yet operating leverage.
Slow thinking still matters
There are founder decisions I would not want AI to make quickly.
Hiring. Firing. When to push. When to wait. Whether a customer is asking for something strategically important or just urgently distracting. Whether the team is tired because the plan is hard, or because the plan is wrong.
AI can help collect the evidence. It can help frame the options. It can help show the consequences of each path. But it should not erase the pause.
The pause is where values enter the decision.
That is the part of work we need to protect as AI gets faster. Not by rejecting AI, but by designing collaboration patterns that keep humans in the loop in a meaningful way.
What I would tell my 24-year-old self
I would tell him Gates was right, but incomplete.
The future was not only business moving faster. It was teams needing a better way to think together as information, software, and intelligence all became more abundant.
I would also tell him this:
Speed will impress you. Be careful.
The important question is not how fast you can produce an answer. It is whether the right people can understand it, trust it, improve it, and act on it.
That is where the next operating layer is being built.
Not in another private AI tab.
In shared work, shared context, and shared intelligence.
The point is not faster alone
I still think about that Barnes & Noble floor.
The book mattered because it gave language to a shift that was already happening. Companies were becoming digital. Information was becoming fluid. Work was changing shape.
We are in another shift now.
AI will make many things faster. That part is obvious. The more important question is whether it will make teams more coherent.
Can people and AI agents work in the same context? Can the history stay visible? Can the output become durable? Can action stay bounded and approval-visible? Can a team think better together instead of each person racing ahead alone?
That is the future I care about building.
Not business at the speed of thought.
Thinking better, together.
