Andrew Warner came to AI from accounting. He sees work the way an accountant sees money: if it is not written down in a standard way, nobody but the person who did it can check it, hand it on or trust it. The story of AI in business, as we see it, is the story of work moving from the person into a system that keeps the books on it.
It has come in three ages.
The age of chat
In the first age, people talked to AI. You typed a question into a window and a model answered. The answers were often very good, and for many people this was the first time a computer seemed to understand them.
But the person carried everything around the answer. You decided what to ask. You remembered what had been said yesterday. You copied the answer somewhere useful, and you did the next step yourself. When the window closed, the work closed with it. AI was a brilliant adviser with no memory and no hands.
The age of code
In the second age, AI began to write software, and then to run it. Agents could read a codebase, change it, test the change and try again. For the first time a model did work rather than describing work, and the people who build software felt the change before anyone else.
The limit was the session. An agent could do a great deal inside one conversation and almost nothing between them. Each session started from a blank page. What it learned, what it was asked, what it promised to do next: all of it lived in a chat that would end.
The age of factories
The third age, the one we believe is beginning, is the age of factories. A factory is a unit of a business set up so that people and AI agents run its work together, every day. It names its job, the person accountable for it, who it serves, the records it owns and what it is allowed to do.
Its work does not live in a chat. It arrives through intake, waits in a queue, is decided by a person or done by an agent within the rules the business has set, and leaves a record. What the factory learns becomes knowledge the next round can use. Software does the exact work, AI offers judgment, and people keep the decisions.
Why the third age matters to a business owner
The first two ages helped individuals. The third changes how a business runs, because a factory is the business's own operations, written down well enough for agents to carry them out. That is the source of its power and of its risk.
Its power: we believe that anyone starting a business should be able to begin from pre-built factories, blueprints made to one open standard, and adapt them into the factories that run their own business. We do not promise any result from that. We think it is the right direction, and we are working on the standard that would make it possible.
Its risk: a fully realised factory is a copy of how your business works. Whoever holds it holds your edge. That is why we believe a business should own its factories, on hardware it owns or controls, with AI models plugged in as service providers. We explain why in a separate article.
What the third age needs
Business already relies on books kept in a standard way, which anyone trained can read. Factories need the same, in the harness layer, the layer that connects AI to a business's own work and records. What every factory declares, how every act is recorded, how a business leaves with everything it built. Andrew calls this intelligent bookkeeping. Defining it well, sharing it widely and teaching it is the first stage of our plan.
Infinity OS, the system these ideas are built into, is being built and is not yet released. Nothing in this article describes a released product.