Build Well, and Build Now: Lessons on AI from Silicon Y’All 2026

I recently returned from Silicon Y’All 2026, held at the Ross Bridge Golf Hotel on the outskirts of Birmingham, Alabama, and organised as always by my friends at Founders Investment Bank. Founders assembles the gathering by invitation only every year, bringing private equity and venture sponsors together with the CEOs of growth companies (around 250 participants in all).

The topic, of course, was AI, and its impact on business in general, on valuations, on business models and on business processes. The discussions that grew out of the presentations were amongst the most mature and thoughtful I have yet participated in. I would be surprised to find a group of business practitioners anywhere in the world more immersed in the practical reality on the ground, whether as operators or as investors.

There was little talk — refreshingly — of utopia or apocalypse, and nobody seemed that interested in fantasizing about the sustainability, or otherwise, of the hyperscalers’ multi-trillion infrastructure investments. 

The focus was almost entirely on business processes. Practitioners from tech companies, and from many decidedly non-tech ones, compared notes on implementing AI professionally, with a clear concentration on ROI. Providers and deployers of private capital alike took the same pragmatic position. AI is here, it’s real, so what does it mean and how do we use it intelligently?

The answer I brought home is that the new AI-driven economy will reward the businesses that build well, and that the window to start is shorter than I thought. Further, neither the building nor the urgency can be delegated. Both belong to the CEO.

The builders will win

If I had to distill the core message of this highly focused crowd into one insight, it would be that successful implementation of AI at the enterprise level (as opposed to the individual level) depends on what every serious business build-out has always depended on: the skilled and disciplined execution of a well-designed engineering plan.

In reality, very few businesses are good builders. But the ones that are will win bigly.

Several of the speakers described what that looks like in practice. Will Blackburn and Peter Inge, drawing on more than 100 AI transformations of the software development process, called what most companies are doing today “random acts of AI”: a coding assistant pilot here, a hackathon there, none of it tied to a roadmap. One of their lessons was simply that not everyone will be a good builder.

Victor Cheng’s prescription for the chief executive was the opposite of random. Pick the one workflow that matters most (“One, not ten”). Commit to a target, such as bringing the timeline down from three days down to 30 seconds, rather than just going for “faster.” Then, as he puts it, “don’t delegate the finish.” See it through personally rather than handing it off once the work is under way.

Jason Parrish argued that good building starts before anything is built, with writing down how a particular task is actually done. His example: “Build an approval workflow.” Inside that simple instruction sit questions any finance director would recognise. Who can approve? Above what amount? Who covers when that person is away? Can someone approve their own request? A good builder settles those questions first. A poor one hands the instruction to AI and lets the machine decide — and it will, filling each gap with an assumption. In Parrish’s words, “An assumption can become software before anyone realizes it was a decision.” 

None of this is new. It is the old discipline of building things properly, applied to a new and very powerful tool.

From the CEO’s desk to the team

“Systems thinking” was a theme woven through all the presentations and conversations, and it matters for something I keep seeing in my advisory work. I have been pleasantly surprised by how many of the owners and CEOs I engage with are already experimenting personally with AI — mostly Claude but also ChatGPT and occasionally Grok — and embedding LLM support into their own desk work.

That is a good start. The overarching need, though, is to move AI usage from personal productivity to group, team or divisional productivity. That’s an entirely different proposition, and one which requires real systems engineering skills to build well.

For owners weighing that move, I recommend a recent essay by the Substack writer Wyndo, “Why Systems Thinkers Are Better at Using AI.” It offers a useful framework for evaluating your own process and professional implementation.

The window of opportunity is closing quickly in the era of AI arbitrage

In April, I published an essay entitled “The Bridge to Wealth in a Ghost Economy.” Its argument ran roughly as follows:

AI has opened a window of arbitrage for the businesses that adopt it early. That window is open now, and it is closing. Only those who move early can exploit the opportunity long enough to build a durable advantage. The early adopters themselves will be the agents of destruction, eroding the premium their competitors have long enjoyed. The chasm between the few and the many will widen until it destroys the businesses left on the wrong side of it. And the firm of tomorrow will have far fewer employees.

My advice to owners was to capitalise on their deep domain knowledge, to lock down their positioning in their clients’ minds while they still can, and never to lose sight of the fundamentals — above all an extraordinary client experience.

After the conference, we asked our own Claude-based AI environment to analyse the presentations I found most useful over the two days and map their main findings against that essay’s claims. I wanted to know whether the thesis had weathered the last half year.

It appears that it has, but the timeline has been substantially shortened. I had given early adopters a window of opportunity of five to seven years to exploit. The consensus now looks more like 18 months to three years.

The bridge to wealth is still being pulled up. Only it’s happening faster than I thought.

The CEO’s job

Every major thread from Silicon Y’all led back to the same desk: that of the CEO. The discipline of building well has to come from the top, which is why Cheng told chief executives not to delegate the finish. Moving AI from the owner’s desk to the whole team is a change only the owner can drive. And with the window closing faster than I thought, there is less time to wait for someone else to start.

Zane Tarence, Senior Partner and Co-Founder of Founders IB and our host for the conference, put it more bluntly. If you as a CEO are not on top of AI and driving the change, you have probably outlived your usefulness.

Hard words. But given the speed of AI and its substantial penetration of all businesses with knowledge work at their core (which is to say, all of them), probably fair.

Further reading from the presenters

Two of the Silicon Y’All presenters have new books on AI out now, and both are worth reading.

• Victor Cheng, Teach AI Your Company: How to Turn Knowledge Trapped in Your Company into Enterprise Value
• Shiv Narayanan, AI Marketing Blueprint