Paul Eremenko's latest company does not make a plane, a phone, or a rocket. It makes the thing that designs them. P-1 AI, which he co-founded and runs as CEO, came out of stealth in April 2025 with $23 million in seed funding led by Radical Ventures. Its first product has a name, a personality, and a job description: Archie, an AI agent built to do the cognitive work that human engineers do when they design physical systems.
Archie is not a chatbot bolted onto a CAD program. The pitch is that he shows up inside Slack or Microsoft Teams like a remote teammate, a junior engineer parked at an offshore engineering center, someone you can task and then review. He starts with the dull and repetitive work. He learns from feedback. And, if Eremenko is right, he gets better the way a young engineer gets better, except faster and without forgetting.
The bet underneath it is unusual. Most of the AI boom has been spent on words and pictures. Eremenko is aiming the same machinery at brackets, ducts, wiring harnesses and load paths. The hard part is data. There is no internet-sized corpus of how a thermal management system gets designed, so P-1 AI generates synthetic engineering data to teach Archie what good looks like. The co-founder handling much of that is Aleksa Gordic, formerly of Google DeepMind and Microsoft.
The early goal is deliberately modest, which is itself a tell. Eremenko does not promise a genius. He promises a competent junior who makes the kind of mistakes a junior makes, inside engineering organizations that already have checks and balances to catch them. The ceiling, though, is not modest at all.
