How To Build A Company With AI From The Ground Up

A new paradigm is emerging for the highest-velocity companies: AI software factories. If you're familiar with test-driven development, this is its next evolution. Humans define what to build by writing a spec and a set of tests that define success. AI agents then generate the implementation and code, iterating until the tests pass. The human's job is to define the outcome and judge the output — the actual code is the agent's responsibility.

Some companies have pushed this to the extreme. Their repositories contain no handwritten code, just specs and test harnesses. Strong DM's AI team is a prime example: they built a software factory where specs and scenario-based validations drive agents to write tests and iterate on code until it meets a probabilistic satisfaction threshold. Their end goal was a system that eliminated the need for a human to write or review code — and it works.

This approach enables what has been called the «thousand-fold engineer» or even the «ten-thousand-fold engineer». A single engineer surrounded by a system of agents can now build things that would have been impossible for an entire team in the pre-AI era. The bottleneck shifts from writing code to defining the right problems and tests.

The critical shift for AI-native companies is maximizing token usage, not headcount. One person equipped with AI tools can now accomplish what used to require a large engineering team. That means dramatically leaner engineering, design, HR, and admin functions. You should be willing to run an uncomfortably high API bill because it replaces what would have cost far more in salaries, onboarding, and coordination overhead.

Jack Dorsey's experience at Block reinforces this. After deep-diving on the tools, he concluded that keeping the same org chart and management structure means missing the shift entirely. The company has to be rebuilt as an intelligence layer, with humans at the edge guiding it rather than routing information through layers of middleware. Every layer of human routing you remove is a direct speed gain — your company's velocity is only as fast as its information flow.

If you're an early-stage founder, you have a massive structural advantage. You don't have legacy systems, entrenched org charts, or thousands of employees to retrain. You're small enough to design your company correctly from day one — workflows, culture, and systems all built around AI as the operating system. This allows you to operate at a speed that is orders of magnitude faster than incumbents.

Existing companies face the opposite challenge. They must maintain and grow live products while unwinding years of standard operating procedures and assumptions about how software is built. Some succeed by spinning up small internal skunkworks teams — Mutiny is a strong example — but for most, every change to core processes risks breaking something that already works. Startups don't have that constraint, and that's an edge to exploit ruthlessly. You cannot outsource your conviction on these tools. Sit with coding agents, use them until you break your own priors about what's possible, and build your company around that new reality.