Founder Story
The company computer
We stopped explaining AEQI in the words of the field and started calling it what it is: a computer for your company. One server, ten officers already in their seats, one memory, one inbox. Here is what changed, and why.
September 14, 2026 · Luca Eich
For a year I described AEQI in the words of the field. Programmable companies. An operating system for the agent economy. Autonomous execution. Every one of those phrases was true, and none of them told a stranger what they would get. People nodded and asked what it did.
So we stopped. AEQI is a company computer: one machine that runs your company. It sits on a server of its own in Germany. Ten officers are already in their seats when it boots — a CEO, a CTO, a COO, a CFO, a CMO, a CRO, a CPO, a Chief of Staff, a General Counsel and a Founder’s Associate. They share one memory. They answer one inbox, across email, Slack, Teams, Telegram and WhatsApp. They open work on a board, and they ask you before anything is published, sent or paid.
That is the whole description. It fits in a sentence because the thing itself is one object. The personal computer gave a person a machine of their own; the company computer gives a company one.
What changed
The runtime did not. It is the same one that has run our own company since June — every quest, decision and line of code behind aeqi.ai was shipped by its officers, and the counts on the front page are read live from that ledger. What changed is how the product presents itself, and three things in it.
- Ten officers, preinstalled. A new Company no longer starts empty. The ten seats exist from the first minute, each with its jobs, its tools and its skills; the CEO writes the week’s plan before you have asked for one. Nothing below the officers is preinstalled: a Head of Design or a Head of Data is hired when the work earns it, as a decision the CEO puts in front of you. A human can take any seat.
- One model row. Every officer runs on the model its work needs — Anthropic, OpenAI, Google, DeepSeek, Mistral, Meta, Qwen, Kimi, Alibaba, Z.ai — set per seat and switched any time. Every model, one company.
- A front page that says it. The eyebrow reads “The Company Computer”. The team is shown as it is in the app, ten coloured orbs with a register you can read. The computer’s parts — memory, agents, server — are drawn, not described. The proof is one live number.
Why the words matter
A name that has to be taught costs you the first sentence of every conversation for years. A name that explains itself gives that sentence back. “Company computer” is not a slogan; it is the category we are in, said plainly, and it is literally what you get: a dedicated machine, not a seat in someone else’s.
It also draws the line we care about. Most of what is sold as an AI employee is a very good tool inside your existing tools. A company computer is the other way round: the company is the object, the agents are its staff, and your tools plug into it. When a model gets better next month, you switch a seat. The company keeps its memory.
What stays
The rules that made the first version trustworthy stay in force. Agents ask before anything that matters leaves. Nothing is reported done that nobody has seen working. Your data stays on your machine, in the EU. Plans stay at €50, €200 and €800 a month with a three-day trial, and the founding offer runs until 31 October.
If you have read our earlier essays — the Company is the primitive, why AEQI pivoted — nothing in them is retracted. They explain the machine. This post is the name on the box.