Ex-Palantir founders raise $30M+ to build the knowledge layer AI agents are missing
What's the deal? Edra, a New York and London-based enterprise AI startup, has emerged from stealth with more than $30 million in funding led by Sequoia Capital, 8VC, and A*Dealroom has a profile for this one. Try Dealroom →. The company was founded by two former Palantir engineers who spent years embedded inside large organisations building AI-driven automation systems. Its platform captures how businesses actually operate — not how they document they operate — and turns that into executable instructions for AI agents.
Customers including ASOS, HubSpot, and Ergeon are already live on the platform, with Edra claiming it can get new customers up and running in as little as a week.
Why now? Enterprise AI deployment is hitting a consistent wall: models are capable, but they lack the operational context to be useful inside specific organisations. Real processes, exceptions, and decisions live in employees' heads, not in documentation. As companies rush to deploy AI agents across IT service management, compliance, customer support, and more, the absence of reliable, up-to-date operational knowledge is becoming the primary bottleneck.
Edra's founders saw this pattern repeatedly at Palantir. Their argument is that "pointing an agent to a stale PDF in SharePoint" is becoming as outdated as using a physical map for navigation — and that the market for solving this problem is large and largely unaddressed.
What could go wrong? The company's core product — synthesising scattered organisational knowledge into structured, AI-readable instructions — is technically ambitious and hard to validate from the outside. The claim that its system reconciles conflicts, eliminates outdated practices, and improves continuously with every execution is compelling, but enterprise software is littered with tools that worked brilliantly in pilots and stalled in production.
Adoption also requires organisations to trust a third party with sensitive operational data — tickets, emails, logs, and internal processes. That is a meaningful barrier in regulated industries, even with early customer traction.
The signal: Edra is betting on a category it calls "executable knowledge" — a layer between raw data and AI agents that translates how a business actually works into something agents can reliably act on. The analogy to GitHub is instructive: just as code collaboration needed a shared, auditable platform to scale, operational knowledge for AI agents needs the same infrastructure. If that framing proves correct, the market is large. AI labs are already paying companies like Scale AI hundreds of millions to generate training data — Edra is positioning itself as the enterprise equivalent, but for operational intelligence rather than model training.
A.M.
