Seattle startup Endeem has raised a $2.5 million pre-seed round and made its AI agent control platform generally available, giving enterprises visibility and rules over agent activity.
Key Takeaways
- Endeem closed an oversubscribed $2.5M pre-seed round, co-led by Beta Boom and Auriga Cyber Ventures.
- The platform became generally available on November 17, 2026, after pilots in pharma, finance and IT software.
- Valuation and pricing are undisclosed.
Lead
Endeem, a Seattle-based startup, emerged from stealth on November 17, 2026 with a platform that assigns each AI agent its own identity, records what it does and what it costs, and checks every request against company-defined rules. The company also announced a $2.5 million pre-seed round co-led by Beta Boom and Auriga Cyber Ventures. Treeo VC and strategic angels also participated. Endeem did not disclose its valuation.
What Does Endeem Actually Sell?
Endeem sells a control layer that sits between AI agents and the systems they touch. Each agent receives an individual identity, so security teams can separate agent activity from employee activity. The platform shows what agents did, what prompts they ran and what they cost.
Policy enforcement covers three connection types: APIs, Model Context Protocol (MCP) connections and command-line interfaces. Companies can allow routine tasks automatically, require human approval for sensitive ones and block requests that break the rules. Instead of permanent access, agents receive short-lived credentials issued per request.
The product works with existing tools, including Claude Code, Codex, Claude Cowork and Copilot, as well as custom-built agents. It runs on employee laptops and in cloud environments, and Endeem says agents need no modification to be covered.
Why Do AI Agents Need Their Own Identities?
Most agents today run on the credentials of the employee who launched them, which leaves security teams unable to tell who or what took an action. Cost attribution suffers for the same reason, since agent spending blends into a human user's footprint.
Founder and CEO Yannick Pobiega framed the problem in the company's announcement: "When a company hires someone, nobody hands them a colleague's password, yet that is how most AI agents work today." The comparison is blunt, and it matches how many early agent deployments were wired up: fast to start, loosely governed.
Identity is the foundation for everything else Endeem offers. Without a distinct agent identity, an audit log cannot say whether a database query came from a person or a script acting on that person's behalf, and a rule cannot be written for one but not the other.
How Big Is the Round, and What Does It Signal?
At $2.5 million, the round is small. Pre-seed rounds in enterprise security often run larger, and Endeem has not said how many months of runway the money buys or how much of it is earmarked for hiring. The "oversubscribed" label is the company's description; the number of investors turned away is not public.
The investor mix is more informative than the check size. Auriga Cyber Ventures is a security-focused backer, which places Endeem in the cybersecurity budget line rather than the general AI tooling line. That distinction matters for sales, since security teams typically hold veto power over software that touches credentials.
The company has no disclosed revenue, and pricing is undisclosed. The traction it points to is pilots rather than contracts.
Who Is Testing It?
Endeem says it is running pilots with enterprises in pharmaceuticals, financial services and IT software, including a top 20 global pharmaceutical company. None of the customers is named. Regulated industries are a logical starting point, because compliance teams there already demand audit trails for human access and are beginning to ask the same of automated actors.
Pilots are not production contracts, and the gap between the two is where many early security products stall. Procurement at a large pharmaceutical firm can run for quarters, and a pilot rarely guarantees expansion.
Who Does Endeem Compete With?
Endeem enters a field that includes established identity and access management vendors, cloud security platforms and a growing number of agent-specific startups. Incumbents are adding agent features to existing products, which gives them distribution that a pre-seed company lacks. Startups focused on agents argue that human-oriented identity tools were not designed for software that acts autonomously, spawns tasks and holds credentials for seconds rather than years.
Endeem's bet is that a dedicated layer, working across multiple agent vendors, will beat features bundled into any single vendor's stack. Support for tools from competing AI providers is consistent with that position. Whether buyers want a separate control plane or prefer to extend what they already own remains untested at this stage.
What Comes Next for Endeem?
The near-term test is converting pilots into paid deployments and naming reference customers. Pricing will also matter: usage-based models tied to agent activity would track how fast customers expand their agent fleets, while per-seat models would not.
A follow-on seed round is the likely next financing event, and its size will show whether the pilots turned into revenue. The broader market is moving in Endeem's favor, as enterprises that spent 2025 and 2026 experimenting with coding and productivity agents now face questions from auditors and security leaders about what those agents can reach.
Outlook
Endeem has a clear product thesis, a security-focused investor base and early enterprise pilots, but it also has a small round, no disclosed revenue and larger rivals closing in on the same problem. The next twelve months will show whether agent identity becomes its own product category or a feature absorbed by existing security vendors.



