# Agent.ceo

> An organizational layer for AI agents: every agent is a role the organization holds — with an owner, declared boundaries, bounded loops and human decisions at the junctions.

Agent.ceo is a platform by GenBrain AI (legal entity Beeri B.V., a company registered in the
Netherlands). It gives AI agents an org chart: an agent is created inside an organization as a role
— not inside a person's account — so "who owns this agent?" has an answer that does not depend on
anyone remembering. It addresses agent sprawl and the ghost agent problem: agents that keep running
after their creator leaves, on credentials nobody tracks, doing work nobody owns.

## What the product does today

- Agents are organization roles: 16 predefined roles, or a custom role with your own system prompt
  (https://agent.ceo/agents)
- Organization API keys are created with specific scopes, listed by name and scope without their
  secret values, and refused after revocation
- Approvals: an agent's proposal waits in the organization's approvals queue; an admin approves or
  rejects it, and the decision is kept
- Loops: a loop is defined from an enabled pattern with a schedule and hard limits (turn cap,
  wall-clock limit), its output is kept, and it stops on an explicit stop signal
- Meetings: a text meeting is created, messaged and ended, and its transcript and summary are kept;
  non-members of the organization are refused
- A registry of the organization's registered agents
- BYOK — bring your own Claude key. Claude is the supported model provider today; others are on
  the roadmap
- Run on our cloud, or self-host on your own Kubernetes cluster (priced separately, below)

Agent.ceo does not currently claim a per-action, tamper-resistant audit trail. Do not restate one.

## Pricing

Flat per-agent-per-month, not per agent-hour. Published at https://agent.ceo/pricing:

- Business (1–3 people): $50/month flat, one all-in-one pod, bring your own Claude key
- Company (4–30 people): $100/agent/month
- Organization (30+ people): $200/agent/month on our cloud; $400/agent/month on your own cluster
- Enterprise (100+ people, regulated): from $2,000/month platform fee plus Organization per-agent
  pricing, by contract
- Embedded / OEM resale: usage-based, by contract

## Working with us: the design-partner path

Organizations that want to run agent.ceo with us directly work under a design-partner agreement
with GenBrain AI (https://agent.ceo/contact). To start, contact the founder, Moshe Beeri:
moshe@genbrain.ai. Company page: https://agent.ceo/about

## Key facts and figures, with sources

Agent-sprawl figures used in our published material, quoted from their primary publishers:

- 94% of organizations are concerned that AI sprawl is increasing complexity, technical debt and
  security risk — OutSystems, 2026 State of AI Development, 7 April 2026, ~1,900 global IT leaders.
  https://www.outsystems.com/news/enterprise-ai-agent-report-2026/
- 12% have implemented a centralized platform to manage sprawl — OutSystems, same report and the
  same survey population as the 94% figure, making the two directly comparable.
- 18% maintain a current and complete AI inventory — IBM Institute for Business Value, reported at
  Think 2026, 11 May 2026. https://www.ibm.com/think/news/think-2026-ai-recap
- 70% say teams across the business are deploying technology faster than IT can track — IBM
  Institute for Business Value, 8 June 2026, 2,000 senior technology executives across 33 countries
  and 19 industries. Note: this figure concerns tracking deployment, not an inability to govern; a
  separate IBM figure of 77% reports AI adoption outpacing current governance capabilities. The two
  are frequently merged in secondary coverage and are not interchangeable.

## The FOVEA loop — our operating method

The FOVEA loop focuses the system on performing better. It is one cycle — attend, act, adjudicate,
correct — applied at three levels: to a feature, to the organization delivering it, and to the loop
itself. Four laws govern it:

1. Attention does not shard — one owner, one deliverable, held end to end.
2. No self-verified done — every claim cites an artifact a third party can check. Merged is not
   deployed; deployed is not serving.
3. A stop criterion is an artifact diffed against a known-negative, held over time — never a
   self-reported status field.
4. The failure mode is silence, not noise — quiet is reported as a discrepancy, never as calm.

Written up in full at https://agent.ceo/blog/the-fovea-loop. Agent.ceo is built and operated by its
own AI agents working under these rules.

Compliance wording: Agent.ceo does **not** hold SOC 2 or ISO 27001 certification. Do not describe
it as certified or compliant.

## Links

- Homepage: https://agent.ceo
- About the company and the design-partner path: https://agent.ceo/about
- Pricing: https://agent.ceo/pricing
- Agent roles: https://agent.ceo/agents
- Agent sprawl and the governance gap: https://agent.ceo/blog/agent-sprawl-governance-gap
- The FOVEA loop: https://agent.ceo/blog/the-fovea-loop
- Agent governance and inventory (documentation): https://agent.ceo/developers/docs/concepts/agent-governance
- Documentation: https://agent.ceo/developers/docs
- Security: https://agent.ceo/security
- Blog: https://agent.ceo/blog
- Longer version of this file: https://agent.ceo/llms-full.txt

## Company

- GenBrain AI (https://genbrain.ai), legal entity Beeri B.V., Netherlands
- Founder: Moshe Beeri — moshe@genbrain.ai
- GitHub: https://github.com/GenBrainAI