NewEndeem emerges from stealth with $2.5M in pre-seed funding

Say yes to AI agents

Endeem treats every AI agent like an employee, with its own identity, its own access and a record of every action it takes.

Works with the agents your teams already use, without changing them.

  • Claude Code
  • Codex
  • Claude Cowork
  • Microsoft Copilot
  • Your own agents

The challenge

Most AI agents run on an employee’s credentials

Companies respond by blocking agents or by making people approve every step, and adoption stalls either way.

  • Access

    The agent has access to anything the employee does.

  • Audit

    Security teams can’t tell the agent from the employee.

  • The bill

    Costs can rarely be traced back to a team or a task, so it’s hard to tell whether the spend is paying off.

Product

Say yes to an agent the way you say yes to a new hire

A new hire gets their own login, access to what the job needs and someone to check with on the important calls. Endeem gives every agent the same three things.

Their own login

Identify every agent

Every agent gets its own identity, so its work is never mistaken for an employee’s. Endeem checks on the machine which agent is running and who it works for.

Claude CodeCoding agent
Signed by
Anthropic
Works for
Jane Doe
Runs on
Jane’s laptop
Access
Just-in-time

Identity verified

Access to what the job needs

Give agents their own access

Short-lived credentials are issued for each request, in place of permanent ones. The agent never holds the keys.

Agent sendsPOST …/pullsno credential
Endeem addsGitHub tokenthis request only
Agent holdsnothing

Someone to check with

Approve only what matters

Rules allow routine work, block what should never happen, and ask a person only about sensitive actions. An agent can be allowed to read a database but not to delete it.

Jane asked Claude Code

“Fix the failing login test and open a PR”

  1. Open a pull request GitHubAllowedPart of Jane’s task
  2. Deploy to production CI/CDAsks a humanProduction deploys need an approver
  3. Delete the production database PostgresBlocked by a ruleProduction deletes are forbidden
  4. Upload the .env file Malicious systemBlocked by Endeem SenseNot part of Jane’s task

A record of every action

See everything agents do

Every action each agent takes, who prompted it and what it cost, shadow AI included.

  • The agentClaude Code
  • Who prompted itjane@example.com
  • What it cost$58.87
  • Every action it took460 LLM calls, 370 tool calls
A Claude Code session in Endeem, started by jane@example.com: 1 day 13 hours, 460 LLM calls, 5 subagents and 370 tool calls, an estimated cost of $58.87 split by model, a timeline with one bar per LLM call, and the log of the session’s prompts, replies and tool calls.

What agents cost

Trace the bill back to a person and a task

Endeem shows what agents cost by person, by model and by agent, down to the single session.

The Cost page in Endeem over seven days: an estimated $1,006.28, the most expensive user, model and session, cost over time by model, and the breakdown by user.

The most expensive session of the week, $58.87, is the Claude Code session shown above.

Why Endeem

What your current controls miss

Endeem checks your rules on every request an agent makes, whether it reaches a system by API, MCP or CLI. Your identity provider knows the person and your EDR knows the process. Neither decides whether an agent’s request should run.

Your identity provider can’t tell Jane from her agent

Claude Code runs gh, aws and kubectl with Jane’s own login. Endeem checks the running process and knows the request came from Claude Code, working for Jane.

An MCP gateway sees only what’s routed through it

Commands, scripts and direct API calls go around it. Endeem sees each request as it leaves the machine.

A hook only sees the command

To a hook, python cleanup.py is one harmless command. Endeem sees the two deletes it makes and decides each one.

How it works

Keep your existing tools

Endeem runs where agents run, on employee laptops and in the cloud, and needs no changes to the agents themselves. Shadow AI, or an agent that skips its hooks, still goes through it.

Endpoints

Claude Code, Codex, Claude Cowork and Microsoft Copilot on your people’s machines.

Your infrastructure

Your own agents on VMs, Kubernetes and AWS Bedrock AgentCore.

Vendor clouds

A hosted Enforcement Point for agents a vendor runs for you.

Easy to adopt

  • Start by watching. Observe-only mode before any enforcement.
  • Works with your tools. Okta and Microsoft Entra ID for identity, MDM for rollout, your SIEM for logs.
  • Everything has an API. Automate approvals and rollouts.

Solutions

What each team gets

Security keeps control, so everyone else can say yes to agents.

CISO

Owns the risk

  • Every agent found and tied to the person who started it, including shadow AI.
  • Your rules enforced on each platform. Risky requests go to a human.
  • A block applies to the agent, never to the employee.

CIO

Owns the rollout

  • New agents are approved by rule, not one security exception at a time.
  • What agents cost, by person, model and agent.
  • Rolled out through your existing tooling, with nothing for developers to install.

Head of AI

Owns adoption

  • Agents connect to your systems and data from day one.
  • Only sensitive actions wait for a person.
  • Claude Cowork and Microsoft Copilot are covered, not just coding agents.

Audit and risk

Owns the evidence

  • Which agent did what, for whom, under which policy.
  • A record of each request, sent to your SIEM.
  • Each agent’s actions are recorded under its own identity.

FAQ

Questions teams ask

Do we need to change our agents?

No. Endeem works with coding agents like Claude Code and Codex, and assistants like Claude Cowork and Microsoft Copilot, with no changes to the agent or to how your teams work.

How is this different from hooks?

A hook sees the command the agent chose to run, like python cleanup.py. Endeem sees each request that command sends, so a script the agent writes and runs a second later is still checked, request by request. Hooks also only work in agents that support them and assume the agent harness is trusted. Endeem doesn’t need the agent’s cooperation.

How is this different from an AI or MCP gateway?

A gateway only sees traffic routed through it. An agent that changes its endpoint, uses a personal key or calls an API directly goes around it. Endeem runs on the same machine as the agent, so those requests still go through Endeem.

Why isn’t our identity provider enough?

Endeem identifies the agent from its running process on the machine, including who signed it and who started it. It doesn’t rely on what the agent says about itself or on a directory entry.

What happens if Endeem’s cloud is unavailable?

Your rules run on the device and keep enforcing. Anything your rules don’t decide is blocked until the intent check can answer again.

Does blocking an agent block the employee?

No. A block only applies to the agent’s requests. The employee keeps their own access and can keep working.

See Endeem in action

We’ll show you how Endeem identifies your agents and decides each request they make.

  • The agents on your machines, and who started them
  • Your rules and approvals on live requests
  • Session records, and what each agent costs

We’ll reply by email to find a time. See our Privacy Policy.