P-1 AI

P-1 AI

P-1 AI is building Archie, an agentic AI engineer designed to work as a junior mechanical, electrical, and thermal engineer inside existing hardware teams. Where most AI coding tools target software, P-1 AI aims at physical-world engineering — mechanical, electrical, thermal, fluids, and systems design — with an agent that operates the professional engineering tools human teams already use. Archie is built on a structured representation of engineering designs plus small custom post-trained models, and P-1 AI is focused first on data-center engineering, covering cooling and critical-power systems, with a stated roadmap into automotive and aerospace. It is early, running through design partnerships rather than general availability.

Production credibility: P-1 AI raised a $50M Series A led by NEA, announced in July 2026, following a $23M seed led by Radical Ventures. Its founder and CEO is Paul Eremenko, a former CTO at Airbus and United Technologies, and its board adds Jeff Immelt, the former GE chairman and CEO, via NEA. Angel investors include Jeff Dean of Google DeepMind and Peter Welinder of OpenAI. The product, Archie, is in private preview with design partners rather than a shippable, self-serve tool, so it is catalogued here as an early-stage, high-pedigree entry.

Key Features

• Archie, an agent scoped to act as a junior hardware engineer within human teams
• A structured representation of engineering designs the agent reasons over
• Small custom post-trained models for engineering tasks
• Trained to operate the complex engineering tools human teams already use
• A continual skills-learning capability
• An initial focus on data-center cooling and critical-power design
• A stated roadmap into automotive and aerospace and defense
• Delivery through design-partner engagements rather than self-serve

Ideal Use Case

P-1 AI fits industrial and hardware engineering organizations — starting with data-center infrastructure — that want to offload junior-level mechanical and electrical design tasks to an AI agent. It is best suited to design-partner engagements today rather than teams needing an off-the-shelf tool. It is not usable as a self-serve product yet, so a team wanting an available CAD copilot should look elsewhere for now.

How P-1 AI differentiates

Unlike software-focused coding agents, P-1 AI targets physical-world engineering disciplines — thermal, fluids, electrical, and systems — with domain-specific post-trained models and tool proficiency. It is distinct from in-app CAD copilots by aiming for autonomous agentic engineering work, and its founder and board pedigree from Airbus, United Technologies, and GE anchor its credibility in hard-engineering domains.

FAQ

Q: What is P-1 AI? A: P-1 AI is building Archie, an agentic AI engineer that performs mechanical, electrical, and thermal design work alongside human hardware teams, starting with data-center engineering.

Q: Is it available now? A: Not as a self-serve product. Archie is in private preview, delivered through design-partner engagements rather than general availability.

Q: How is it different from a CAD copilot? A: P-1 AI aims for autonomous agentic engineering work across physical disciplines like thermal and electrical design, using domain-specific models, rather than in-app assistance inside a CAD tool.

Q: Who funds and leads P-1 AI? A: It raised a $50M Series A led by NEA after a $23M Radical Ventures seed. Founder and CEO Paul Eremenko was CTO at Airbus and United Technologies; Jeff Immelt sits on the board.

Q: What is its focus today? A: Data-center engineering — cooling and critical-power systems — with a stated roadmap into automotive and aerospace and defense.

tl;dr

P-1 AI is building Archie, an agentic AI engineer that does mechanical, electrical, and thermal design work alongside human hardware teams, starting with data-center cooling and power. It raised a $50M Series A led by NEA, with a founder from Airbus and United Technologies and Jeff Immelt on the board. Best watched as an early, high-pedigree entry; Archie is in private preview with design partners, not self-serve.