Smallness Was Never the Problem
Nearly every company in Livingston's book started with an embarrassingly narrow version of its eventual product. The founders did not wait for scale to validate their instincts — they shipped something tiny, watched it meet reality, and adjusted. For a founder building with agents today, this discipline is harder to keep because prototyping is so fast. The bottleneck has moved from typing to thinking, which means the risk is not moving too slowly but moving confidently in the wrong direction without noticing.
Judgment Was Always the Actual Job
What separated the founders Livingston documented was not technical superiority but the quality of their decisions under uncertainty — what to cut, who to serve, when to hold the line. That judgment cannot be delegated to an agent. Agents need clear tasks, good context, and defined boundaries before they produce reliable work, and someone has to set all of that. The founder is still the one deciding what is worth building and for whom.
Adaptability Compounds Differently Now
The pivots in Livingston's book happened over months. With agents compressing prototyping to hours, adaptability can now compound inside a single week. But speed without feedback loops produces noise. An AI-native company has to build evaluation into how it works from the start — not bolted on later — so that fast iteration produces signal a human can actually read and act on.
Distribution Was Hard Then and Stays Hard
Not one company in Founders at Work solved distribution through cleverness alone. They earned customer trust slowly. That has not changed. When software is cheap to create and agents can generate it continuously, the scarce things are customer access, earned trust, and the taste to know which output is actually good. The agent writes; it does not sell, build relationships, or own accountability when something goes wrong.
