Building Institutional Memory for the Agentic Future
We are an applied AI team building institutional memory for AI agents so companies never lose knowledge. We are passionate engineers, technologists, and YC founders innovating at the intersection of document workflows and human expertise.
The Knowledge Gap
AI agents today rely only on documented data, which represents just a portion of an organization's intelligence. The remaining undocumented "tribal knowledge" is currently under attack: it walks out the door when employees leave or leaks to external AI models.
Capturing What Others Miss
To solve this, we build agents that capture both documented and tribal knowledge to execute complex work. We are starting with document workflows, where the knowledge is richest.
By embedding directly into everyday tools like Excel and other productivity tools where work happens, Rima captures the "decision traces" — the heuristics, exceptions, and judgment calls that live in experts' heads. These traces compound into institutional memory that makes AI agents accurate, trustworthy, and audit-ready.
Our Wedge: Accounting
We are applying this first to the accounting labor market, where a 75% workforce retirement wave is creating a "knowledge extinction" event. As our agents automate high-volume extraction and reconciliation from PDFs and ERPs, they don't just do the work; they remember how it's done.
The Vision
Accounting is our wedge, but institutional memory is the enduring asset. We are building the infrastructure for a future where every company owns its own model, trained on its own data and context. Our goal is to build a world where companies never lose knowledge.