Nessie (YC F25) — The shared context layer for recursive companies

In their words

Knowledge work is being restructured by AI. Strategy happens in Claude. Research happens in ChatGPT. Building happens in Claude Code and Codex. Meetings get recorded, notes get written, agents run for hours and leave behind traces nobody reads. The people defining the next decade of work are already thinking with AI every day. But none of that thinking compounds. Every session starts from zero. The reasoning behind decisions, the context behind strategies, the evolution of ideas across hundreds of conversations - all of it evaporates. Nessie is the infrastructure that makes it permanent. We auto-sync AI conversations and agent traces across every major provider - ChatGPT, Claude, Gemini, Perplexity, Claude Code, Codex, Cursor, OpenClaw, Hermes, and more - alongside meetings, notes, and email from tools like Granola, Read AI, Plaud, and Obsidian. Everything a team produces with AI flows into one place and compounds over time. Nessie turns that raw, heterogeneous data into an ontology: the live state of a company. Not a static schema but a continuously updated record of what the company knows, what it has decided, who owns what, and how its thinking has evolved, kept current as new work flows in. It is queryable from any agent through our CLI, MCP connector, and plugins, shareable across a team with real permissions, and deployable on-prem for organizations that need it. Then the loop closes. Once a company’s state is legible to its agents, every agent run becomes training signal: which skills worked, which failed, which workflows keep repeating. Nessie feeds that signal back as reviewable improvements to the skills, prompts, and processes the team runs on. This is what makes a company recursive: its agents get better from its own history, the people working with them get better too, and the compounding never stops. Where we are going: the AI-native organization runs thousands of agents that are powerful but amnesiac. Nessie is the layer that gives them, and the humans steering them, a shared memory. Science has journals, law has case law, software has git; AI-native knowledge work has had nothing. We are building it. Founded by two Yale CS grads and former Amazon engineers, now a team of three in San Francisco. Backed by BoxGroup, Pioneer Fund, Y Combinator, Precursor, and more. https://nessielabs.com

Founders, titles and bios as the company published them. Photographs are our own copies of the ones on that record, so they do not break when the original link expires.