Why I Chose Startups to Stay Close to the User Problem

while (world) { observe(); infer(); hallucinate?(); self-correct(); } while (world) { observe(); infer(); hallucinate?(); self-correct(); } while (world) { observe(); infer(); hallucinate?(); self-correct(); }

001101 010011 sigil::bagua seed::entropy trace::mechanical awe::human 001101 010011 sigil::bagua seed::entropy trace::mechanical awe::human 001101 010011 sigil::bagua seed::entropy trace::mechanical awe::human

temperature=0.92 top_p=0.95 sampler=stochastic runtime=deterministic user=astonished temperature=0.92 top_p=0.95 sampler=stochastic runtime=deterministic user=astonished temperature=0.92 top_p=0.95 sampler=stochastic runtime=deterministic user=astonished

☰☱☲☳☴☵☶☷ oracle != truth ritual == interface meaning <- interpretation ☰☱☲☳☴☵☶☷ oracle != truth ritual == interface meaning <- interpretation ☰☱☲☳☴☵☶☷ oracle != truth ritual == interface meaning <- interpretation

seed = hash(question + state); noise = sample(temperature); pattern = deterministic(seed, noise); if (human_can_track === false) mark("mystic"); return explainability_gap(pattern); seed = hash(question + state); noise = sample(temperature); pattern = deterministic(seed, noise); if (human_can_track === false) mark("mystic"); return explainability_gap(pattern);

## Not a person, still persuasive - interface implies intention - language implies confidence - user infers agency => design for interpretability ## Not a person, still persuasive - interface implies intention - language implies confidence - user infers agency => design for interpretability

cast.bagua = pickTrigrams(seed); cast.moonBlocks = deriveHexagram(seed); cast.fortuneSticks = burnModel(intensity); cast.scapula = generateCracks(seed); return readable_fiction(cast); cast.bagua = pickTrigrams(seed); cast.moonBlocks = deriveHexagram(seed); cast.fortuneSticks = burnModel(intensity); cast.scapula = generateCracks(seed); return readable_fiction(cast);

{ "observe": true, "decide": constrained, "act": reversible, "measure": behavior, "learn": weekly } { "observe": true, "decide": constrained, "act": reversible, "measure": behavior, "learn": weekly }

def perceivable_randomness(system): return complexity(system) > attention_budget if perceivable_randomness(llm): user.labels_output = "fate" def perceivable_randomness(system): return complexity(system) > attention_budget if perceivable_randomness(llm): user.labels_output = "fate"

[trace] t=02:13 system murmurs [trace] tokens fall like ash [trace] certainty simulated [trace] mechanism remains [trace] human names it chance [trace] t=02:13 system murmurs [trace] tokens fall like ash [trace] certainty simulated [trace] mechanism remains [trace] human names it chance

After IBM, I made a conscious career choice: stay close to early-stage teams.

That decision shaped how I work, what I value, and the kinds of systems I build now.

Quick definitions (plain English)

User proximity: how directly your work is tied to real user pain and feedback.
Agency: how much authority you have to make and ship decisions.
Lean: reduce waste, run small experiments, learn fast, and iterate from evidence.
Product stool: the core trio of design, product management, and engineering.

Why I made the shift

Large companies can teach rigor, process, and quality baselines. I value that part of my IBM experience.

But I realized I personally do my best work when:

problem definition is still open,
user pain is direct and visible,
shipping decisions can happen quickly,
the people doing the work also own the outcomes.

That is why I leaned into startup environments: Esper, SmartHalo, Unyte, and later building Bonzen.

In those contexts, you cannot hide behind layers. You either solve the user problem or you feel the miss immediately.

What startups taught me about leadership

Small teams move quickly when people are smart, independent, and aligned on the problem.

That sounds obvious, but it is rare in practice.

The founders I admire most treat entrepreneurship like a creative discipline: they compose product, market, timing, people, and constraints into a coherent enterprise.

To me, that is creative work at the highest level. Not decorative creativity, but structural creativity.

Because of that, I prefer being close to founders and operators who are still in contact with the problem itself.

My operating model: the 4-legged platform

I think a powerful early-stage platform can often be modeled as four legs:

Design: shape interaction and user trust.
PM: define scope, tradeoffs, and sequencing.
Engineering: implement robustly and ship.
CEO / executive decision maker: commit direction and resource allocation.

I have worked across the first three legs directly: design, PM, and engineering.

That cross-functional fluency is where my leverage comes from. It lets me translate across disciplines without losing velocity.

Why this matters now with AI agents

Agent tooling changes the economics of small teams.

A single high-context operator can now coordinate a larger share of planning, drafting, implementation, QA, and synthesis than was practical a few years ago.

In effect, one person can increasingly run a “4-person platform” if they have:

clear product judgment,
technical auditability,
strong constraints,
and disciplined review loops.

This does not make teams obsolete. It changes where leverage sits.

My goal is to build and guide these systems so they improve real human experience, not just throughput dashboards.

On abundance and distribution

I am optimistic about automation-driven surplus.

History already gives us a partial analogy: in many places, food production is no longer the core technical bottleneck; distribution, access, and incentives are.

AI likely follows a similar curve:

production of useful outputs becomes cheap,
distribution of benefits becomes the real governance problem.

So the question is not “can we produce more?” It is “who benefits, who gets access, and who stays in control?”

That is where product design, market design, and policy design need to meet.

Bottom line

I chose startups because I wanted to stay near the source: the user problem, the shipping loop, and the consequences of decisions.

That choice shaped a working philosophy I still trust:

stay close to reality,
move with high agency,
keep teams small and accountable,
and build systems that return value to human life.

Related notes:

Your human friend,
Oli
June 20, 2024