Edra, a New York-based automation platform, emerged from stealth this week announcing total funding of more than €26 ($30) million. Sequoia Capital led a €20.73 ($23.8) million Series A; an earlier €5.66 ($6.5) million seed round was co-led by 8VC and A*, the fund of serial entrepreneur Kevin Hartz. HubSpot Ventures joined the round alongside Sequoia, adding a strategic dimension given that HubSpot itself operates as an Edra customer.
The Series A was led by Sequoia partner Luciana Lixandru, who frames the investment as a bet on both a specific vertical and a broader platform play: “I think they can go really far by doing [IT Service Management]. Then there’s opportunity to become a horizontal platform in the enterprise.”
How Edra started?
Croatian Eugen Alpeza (CEO) and co-founder, Greek Yannis Karamanlakis, met at university thirteen years ago and spent years at Palantir before launching Edra.
Alpeza led the launch of Palantir’s AI platform and drove large commercial client acquisition, while Karamanlakis was the company’s first Forward Deployed AI Engineer — a discipline Palantir pioneered that embeds engineers directly inside client organizations to personalize technology for each business.
The pattern they observed at Palantir became the founding insight for Edra. “The hardest part of automation was never the AI. It was capturing the knowledge,” Alpeza writes in his X blog.
Companies accumulate vast operational data (emails, support tickets, logs, chat histories) but the real processes, decisions, and exceptions live in people’s heads, not in documentation. Edra says it analyzes that data automatically, builds a knowledge base from it, and keeps it updated. And as Alpeza puts it: “Deploying AI in any large organization [requires that] you have a clear account of how you want things done today. And no large organization actually has that.”
How Edra wants to build executable knowledge
Edra’s largest business line now is IT service management. Customers including ASOS, HubSpot, Cushman & Wakefield, and easyJet run Edra on top of their existing systems. The company says it gets organizations up and running in as little as one week.
Edra connects to systems organizations already use, like ServiceNow, Jira, Zendesk, Salesforce, Outlook, and ingests existing data alongside any available standard operating procedures. It then runs what the company calls agentic learning: thousands of AI agents operating in parallel to explore that data, surfaces gaps and inconsistencies in existing processes, simulate decisions, and synthesize conclusions.
The output is a “white-box library of executable knowledge”, as they frame it — plain-English instructions that AI agents can follow, that domain experts can read and review, and that update continuously as the business changes.
How they achieve this? Edra positions itself against two common but allegedly inadequate approaches.
The first is asking companies to hand over their SOPs: documentation that, in most large organizations, is incomplete, inconsistent, or simply does not exist. The second is treating knowledge capture as a search problem: pointing an agent at historical data every time it needs to act. Alpeza argues this fails because “that data is full of contradictions, outdated guidance, and one-off decisions that were never meant to be permanent policy.” Edra’s learning system tries to reconcile those conflicts and produce auditable, changeable instructions rather than a static archive.
“Edra is fundamentally different. Our learning system synthesizes scattered information into explicit, auditable executable knowledge, reconciling conflicts, eliminating outdated practices, and improving with every execution,” Alpeza claims.
Why they might make it?
All that being said, Edra is not the first nor last to build solution like this (though the approaches vary). Not even half year ago, I had opportunity to speak to Stefan Damm, a seasoned engineer, manager and co-founder of Mimirio, an Austrian startup specializing in knowledge retrieval systems.
What might make Edra the leader of the pack? As always, combination of ambition and execution. It is no wonder Alpeza draws a direct parallel to GitHub: “Think about what GitHub did for code: it gave software teams a shared platform to collaborate openly, with clear ownership, built-in review, and a complete history you can audit. Operational knowledge needs the same shift.”
The duo believes Edra can evolve into infrastructure that every knowledge worker uses to teach AI how their job actually works, at a time when businesses are already spending heavily to supply AI models with domain expertise.
