In a leadership meeting this spring, a teammate admitted she had touched a part of the company's platform that handles payments and invoices. She was trying to be helpful. The founder thanked her for saying it out loud, then drew the boundary in one sentence: payments are a no-go zone, because testing a payment path carries a rigor the rest of the codebase never demands.
What struck me was the speed of the catch. He understood the change within minutes, knew why it mattered, and reset the line without a trace of blame. He could do that because his hands were still in the product. He had built the previous version of several pages himself.
A decade ago, that founder gets labeled a micromanager and coached to let go. The playbook of that era said hire great people and get out of the building. I taught versions of it myself. AI broke it.
TL;DR: Founder proximity means staying close enough to the actual work that your judgment shapes what ships. AI made execution abundant, which moved the founder's edge upstream: the vision itself, and the judgment to catch what the tools miss. The delegate-everything playbook now costs more than it saves, and the founders who stay close are winning.
I've coached 500+ entrepreneurs from 31 countries over 20 years, co-founded CTOx (a $2.4 million accelerator for fractional CTOs), and spent 8.5 years inside Abundance 360 studying technology curves before they went mainstream. Since 2023 I've run my business on a 3.5-day workweek with AI carrying most of the execution, so the proximity question is one I live daily: what stays in my hands, and what goes to the machines.
For 20 years, founder maturity was measured by distance from the work. Prove the concept, hire specialists, work on the business instead of in it (Michael Gerber's phrase outlived his book), and graduate from touching anything a specialist could own. Distance became a status symbol.
Paul Graham's Founder Mode essay named the cost. Brian Chesky had followed the standard advice, hire good people and give them room, and watched Airbnb drift toward results other people wanted. When Chesky told that story to a room of successful founders, one after another said the same thing had happened to them.
I watched a gentler version of the drift. The founders I know who delegated everything are doing fine; their businesses run without them. What thins out is the thumbprint. On one project, I felt an ache in my chest when it became clear the founder was totally checked out from the thing he wanted built. His business kept running, and his fingerprints were gone from it. The growth that remained leaned on referral relationships formed years earlier, and rooms like that eventually go stale: everybody has already collaborated with everybody else, and there are no new fish in the pond.
Execution used to be the scarce thing. Now 78% of organizations use AI somewhere in the business, up from 55% a year earlier. In Y Combinator's winter 2025 batch, a quarter of the startups had codebases that were 95% AI-generated, founded by people fully capable of writing every line themselves. They chose speed and kept their judgment.
In April I dictated a note to myself after that meeting, and one sentence keeps proving itself: with AI, everybody can be pretty competent at almost any given thing in knowledge work. When competence is abundant, it stops deciding outcomes. The edge moves upstream to what the tools can't supply: the quality of the vision, and the proximity of the vision to the final output. That's the missing link.
YC partner Diana Hu described the same gap from the investor side: "You have to have the taste and enough training to know that an LLM is spitting bad stuff or good stuff." Taste comes from contact with the work, at a frequency no quarterly review can supply.
Proximity earns its keep in the gap between what the tools report and what is happening underneath. METR ran a randomized trial with 16 experienced open-source developers: working with AI tools, they took 19% longer to finish their issues while believing the same tools had made them 20% faster. The direction of the error is the striking part. The tools manufactured a feeling of speed.
The pattern repeats at the company level. MIT's NANDA initiative found that 95% of enterprise AI pilots produce no measurable profit-and-loss impact, largely because generic tools never learn the workflow they were dropped into. BetterUp Labs and Stanford's Social Media Lab surveyed 1,150 desk workers and found 40% had received "workslop" in the past month: AI-generated content that reads as finished work and moves nothing forward, costing nearly two hours of cleanup per instance. GitClear's analysis of 153 million changed lines of code found churn on pace to double against its pre-AI baseline, with copy-paste rising while refactoring fell.
Every one of those failures arrives looking finished. The demo works and the dashboard stays green. A founder in real contact with the work catches the drift early, the way the founder in that meeting caught the payment change: in the conversation where it surfaced, with context no monitoring tool holds.
My standard for collaborators changed this spring: if you're not willing to get your hands dirty building your vision, I don't want to help. That sounds harsh until you watch what proximity does for the people around a founder.
The founder from that meeting has never read as a micromanager to me. His hands are on the wheel even when he isn't building every feature; he's close enough to know what his teammates are creating and why. When one of them redesigned a page whose first version he had built himself, he recognized the improvement on sight and said so in front of the whole team. A yes from someone who knows the terrain carries more weight than a yes from someone reading a status report.
The teammate who touched the payment path is still building, bolder than before, because the boundary gave her a map of where boldness is welcome. When you hold a clear perspective on how the thing should feel and function, and the standard of quality it has to clear, the people around you build even better. They feel the importance and the commitment the team is sharing.
The founder has to give the most fucks in the room. Otherwise the vision runs out of gas. I watch the same physics on a book project: a thought leader I work with turned down a ghostwriter flat, because he wants the writing to be his. My job is organizing his ideas so well that he does the best writing of his career. A book only he could have written is the asset, and his proximity is what makes it one.
Chesky, whose Airbnb turnaround started the founder-mode conversation, spent this spring describing what comes next: AI founder mode, which he expects to demand involvement in significantly more detail than the original, because the tools put every part of the company on demand.
Proximity without structure decays into hovering, so give it structure. This audit takes 30 minutes with your leadership team:
If you want the systems side of this discipline, the build-once-run-weekly approach in AI Systems for a Lean Business pairs with this audit: the systems carry the volume, and your proximity holds the bar.
The founders I'd bet on for the next decade treat AI as an engine, one of many engines, and themselves as the reason the engines point anywhere at all. You cannot delegate your vision and your commitment. Those are the new fuel.
Whether proximity scales to a 5,000-person company, I don't know yet; Chesky is running that experiment in public. In the 2-to-50-person businesses where I spend my weeks, the evidence stopped feeling ambiguous to me: the founder who stays close ships work with a thumbprint and catches drift while it is still cheap. Their teams get the context that turns AI into an amplifier. And the vision-execution gap, the distance between what you see and what ships, narrows for the first time in most founders' careers.
Delegation solved a scarcity of execution. That scarcity just ended. The question worth sitting with this quarter: where does your business still carry your fingerprints, and where would nobody notice if you never looked again?
Ready to decide what deserves your hands and what leaves your plate? The Simplicity Protocol helps ambitious professionals do less to achieve more through weekly elimination strategies you can implement in 20 minutes or less. Subscribe and get one proximity-protecting elimination to run each Tuesday.
Founder proximity is staying close enough to your company's actual output that your judgment shapes what ships. It includes reviewing real work product, knowing what your team and your AI tools are producing and why, and keeping direct contact with at least one stream of work you'd stake the brand on. Distance from the org chart is optional; distance from the output is the risk.
Micromanaging controls how people work. Proximity keeps the founder fluent in what the work is, so feedback lands as calibration from someone who knows the terrain. The founder in the story above stayed out of his teammate's process and still caught the boundary issue immediately.
Delegate execution generously. Keep the vision and the no-go zones in your own hands. AI makes competent output abundant, so a business differentiates on the judgment applied to that output, and a founder who delegates judgment along with execution ships work that could have come from anywhere. The delegation question itself moved: it used to ask what a specialist could own, and now it asks what only you can calibrate.
A no-go zone is a place where automation should never touch your business, no matter how capable the tools get. Payment processing is the classic example, because testing it demands unusual rigor. Naming yours out loud gives your team a map of where to build boldly.
Review one real output stream weekly, the way a customer meets it, and measure outcomes instead of trusting the feeling of speed. In METR's trial, experienced developers believed AI made them 20% faster while it made them 19% slower; nobody caught the gap by feel. Direct contact with the work, plus a real measurement, is what exposes it.
The tools average. Given similar prompts, everyone's output converges toward the same competent middle, which is why 40% of desk workers now report receiving polished, empty workslop. The founder's thumbprint, applied up close, is what pulls output off the average.
The world keeps accelerating. The Simplicity Protocol helps ambitious professionals do less to achieve more through weekly elimination strategies you can implement in 20 minutes or less.
Member discussion