Current AI agents are brilliant amateurs — capable in the moment, forgotten by morning. Project Laplace builds agents that accumulate expertise: persistent memory, genuine personality, and domain mastery that compounds across years, not sessions.
A real employee gets better every year they work with you. They understand the institution, the blind spots, the unwritten rules. No current AI can do this. We are building the infrastructure to change that.
Project Laplace is not a product company. We are building the foundational substrate for AI that persists — the memory architecture, the reasoning engine, and the agent runtime that makes long-horizon intelligence possible.
Our system operates across four layers: a Memory Substrate that never resets, a Reasoning Engine that deliberates continuously, an Agentic Cohort of specialized collaborators, and a Human Interface built on review primitives rather than chat bubbles.
We work in closed cohorts with a small number of research partners. Each cohort runs for a minimum of twelve months. The agents that emerge from that period carry institutional knowledge no prompt can replicate.
Engineer by profession, builder by obsession. I spend most of my time teaching machines to think, designing AI products, and convincing computers to do things they probably shouldn't. Passionate about AI, startups, and building technology that feels a little magical.
Project Laplace