Funding & VC

Endeem Exits Stealth With $2.5M Pre-Seed to Give AI Agents Their Own Identities

Seattle startup Endeem raised a $2.5M oversubscribed pre-seed from Beta Boom and Auriga to give enterprise AI agents individual identities, permissions and cost tracking.

By Olivia Hart

3 min read

Endeem Raises $2.5 Million To Secure Enterprise AI Agents
Endeem Raises $2.5 Million To Secure Enterprise AI Agentsmastersenaiper / pixabay

What's News

  • Endeem raised $2.5 million in an oversubscribed pre-seed co-led by Beta Boom and Auriga Cyber Ventures
  • The Seattle company launched general availability of its security platform for enterprise AI agents
  • Endeem is piloting with pharma, financial services and enterprise software firms, including a top-20 global pharmaceutical company
  • The platform supports Claude Code, Codex, Claude Cowork, Copilot and custom-built systems
  • It enforces policy per request across APIs, Model Context Protocol connections and command-line interfaces

Endeem has raised $2.5 million in an oversubscribed pre-seed round and launched general availability of its security platform for enterprise AI agents. Beta Boom and Auriga Cyber Ventures co-led the financing, with Treeo VC and strategic angel investors participating. The Seattle-based company will spend the capital on product development and enterprise adoption.

The core problem Endeem targets is structural. Many AI agents today operate through an employee's credentials, which lets an agent perform the same actions as the employee while blurring the line between human and automated activity. It also complicates cost management: an organization can see total AI spending without connecting the expense to a specific agent, department or assignment.

How does the identity model work?

Endeem assigns each AI agent its own organizational identity, similar to how companies provision individual accounts when hiring employees. The platform records who prompted the agent, which actions it performed and how much those actions cost — giving security teams visibility into activity, permissions and operating costs per agent.

Founder and CEO Yannick Pobiega framed the gap in hiring terms: "When a company hires someone, nobody hands them a colleague's password, yet that is how most AI agents work today. A new hire gets their own login, access to what the job needs and someone to check in with on the important calls. We built Endeem so a company can say yes to an agent the way it says yes to a new hire."

The promise is accountability and auditability. The caveat: identity assignment only works if identification stays reliable as agent architectures and AI tools keep changing — a challenge the company itself acknowledges.

What policy controls does the platform enforce?

Endeem evaluates company rules every time an agent sends a request through an application programming interface, a Model Context Protocol connection or a command-line interface. Organizations can set three tiers of response:

  • Permit low-risk routine activity automatically
  • Require human approval for sensitive tasks
  • Block prohibited actions outright

An agent might be authorized to read from a database, for example, without permission to delete records. The platform can also issue short-lived credentials for individual requests instead of permanent credentials that grant broader access.

Endeem runs on employee computers and in cloud environments without requiring changes to the agents themselves. It is designed to work with Claude Code, Codex, Claude Cowork, Copilot and custom-built systems. The platform additionally attempts to discover unapproved "shadow" AI activity, building a fuller inventory of agents operating inside the organization.

Who is already using it?

Endeem is piloting the platform with companies in pharmaceuticals, financial services and enterprise software. One participant is described as a top-20 global pharmaceutical company — though neither the company nor pilot metrics are disclosed, so adoption claims rest on the startup's own characterization.

Regulated businesses may have the strongest need for agent-level controls, since automated systems can touch sensitive information or take actions inside production environments.

Can a pre-seed company win this category?

Kimmy Paluch, Managing Partner at Beta Boom, argued the market is moving fast: "The need to control AI agent access and maintain an auditable history of their actions is clearly exploding, and we believe that the winner in this category will need a unique approach and extensive expertise. Endeem has both, and Yannick, the CEO, struck us as an uncommonly driven and resourceful founder. That's a winning combination that was easy for Beta Boom to get behind."

Investor enthusiasm at pre-seed is cheap; enforcement reliability at enterprise scale is not. Endeem's stated challenge is maintaining dependable identification and policy enforcement as agent architectures, enterprise systems and AI tools continue to evolve.

The company's bet is that its employee-style identity model lets businesses adopt agents without manually approving every routine step — or blocking the technology entirely. Whether that model holds up against fast-changing agent architectures will determine if $2.5 million was the cheapest entry ticket in the category or the first of many raises.

Source: pulse2.com

Share this article:

More from Olivia Hart

Olivia Hart

Show full bio

Staff writer covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.

2,378 articles

Related articles

« Previous articleNext article »