When I Became the Bottleneck.

Too Many Things Worth Building

There were more projects, startups, and possibilities than I could reasonably work on at once. I was building to find answers. Some ideas failed quickly. Others opened doors I hadn't expected. I didn't want to stop at one startup and assume it was the only possibility worth pursuing.

The problem was that every project needed attention. The small decisions, fixes, setup work, and constant maintenance competed with the bigger decisions about what to build next. I had become the bottleneck.

The Response: Build a Builder

AutoBuilder emerged from a straightforward question: what if the work of building could keep moving without requiring me to do every small thing myself?

The idea was a 24/7 building system. AI would take on routine, repeatable work while I focused on direction, experimentation, and the bigger picture. That was the aspiration behind the design, not a claim that all autonomous tasks were already reliably completed without supervision.

As the Tools Changed, So Did the Questions

Early AI was useful for smaller pieces of work. The ambition was larger than the models' reliable abilities. As models and agent tools advanced, it became possible to consider more complex delegation: planning, coding, coordination, testing, and revision—while still needing human oversight and independent verification.

AutoBuilder helped reveal other problems worth investigating: how multiple agents should cooperate, where work should be recorded, how outcomes should be checked, and what an organization should remember from completed and failed attempts.

Failing Quickly Was Part of the Method

The purpose wasn't failure for its own sake. It was to try enough things to learn which ideas were promising, which weren't, and what each attempt taught us. A failed venture might reveal a reusable technical component, an incorrect assumption, or a question that later becomes a different product.

But that only becomes compounding knowledge when it is captured, evaluated, and applied intelligently. Otherwise, moving quickly can mean repeating the same mistakes at higher speed.