New York startup Edra has closed a $30 million Series A led by Sequoia Capital, with participation from 8VC and A* — the firm founded by serial entrepreneur Kevin Hartz. The funding brings Edra out of stealth with a focused bet on one of enterprise AI's most persistent problems: agents that don't know how the business actually runs.

The Problem with General-Purpose AI

When companies drop an AI agent into their environment, it starts from zero. Every company has its own escalation paths, workarounds, and tribal knowledge — accumulated over years and rarely documented anywhere. Getting an agent up to speed typically requires expensive forward-deployed engineers, manual documentation efforts, and consultant hours — work that has to be redone every time a process changes.

Edra's founders know this firsthand. Eugen Alpeza spent seven years at Palantir, where he built out the U.S. commercial go-to-market motion and later led the launch of Palantir's AI Platform. Yannis Karamanlakis was Palantir's first Forward Deployed AI Engineer, leading teams focused on taking LLMs from demos into production at scale. The two met at university 13 years ago and always planned to start a company together.

Living Knowledge, Automatically

Instead of asking humans to write documentation, Edra analyzes data companies already generate — support tickets, emails, logs, chat histories — and builds a structured knowledge base from it. Crucially, the system is transparent and editable: you can see exactly what Edra has learned and why. As new data flows in, the knowledge base updates itself.

Current use cases are centered on IT service management and customer support. Early customers include HubSpot, ASOS, Cushman & Wakefield, and easyJet.

Sequoia framed the investment as a bet on a critical infrastructure layer for the agentic era: agents are only as useful as the context they operate with, and Edra's approach — deriving that context automatically from operational data — addresses why so many enterprise AI deployments stall after the demo.