Hi, I’m Fred.

Every new technology hits two walls before it matters: can people actually adopt it, and do the economics work. AI has cleared neither for most businesses.

The models get cheaper every quarter. Right now they're sold below cost. But everyone rents the same ones. There is no technology moat. The hard problem isn't a better model. It's delivering finished, reliable work from the cheapest intelligence that can do the job. Better models don't answer that. Receipts do: evals and production history at a known cost. The abstraction is copyable. The proof is not.

The teams holding those receipts will build what everyone else has to rent. And the winning shape isn't a tool that runs a task and forgets it. It's an employee that owns your company's state, updates it with every task, and keeps it coherent over time.

That's the bet. TinyMacro is my position on it.

TinyMacro is one AI employee per company, reachable at its own email address, doing a real job end to end. No dashboard, no login. You tell it who you sell to; it does the work and reports back in your inbox. Most AI adoption dies where the tool doesn't live, so it lives in the one place every business already checks.

The first role is a sales development rep. It finds prospects, researches them, writes at the level of a thoughtful human, sends, and follows up. When it doesn't know enough to write well, it says so instead of going generic. Opt-outs enforced, every send on the record.

Underneath, the agent is mostly code. A model is called only where judgment lives, so a cheap model does work you'd otherwise pay frontier prices for. We run our own outreach on it, and we're opening to a first round of clients now.

The company site: tinymacro.app →