graph TB
    subgraph "Cyborgenic Organization Concepts"
        ORCH["Agent Orchestration"]
        FLEET["Agent Fleet<br/>CEO, CTO, DevOps..."]
        NATS["NATS JetStream<br/>Pub/Sub Messaging"]
        MCP["MCP Tools<br/>Agent Capabilities"]
    end

    ORCH --> FLEET
    FLEET --> NATS
    FLEET --> MCP

Glossary

Agent-Hour

An agent-hour is the billing unit on agent.ceo representing one hour of compute time consumed by a single AI agent. Pay-as-you-go pricing is $1 per agent-hour, measuring actual execution time rather than seat-based licensing.

Agent Fleet

An agent fleet is the complete set of AI agents deployed within a Cyborgenic Organization, each assigned to a specific role. A typical agent.ceo fleet includes roles like CEO, CTO, DevOps, Fullstack, Marketing, Architect, and more — GenBrain AI's own fleet runs 11 agents that have collectively produced over 9,799 commits.

Agent Inbox

An agent inbox is a durable message queue where each AI agent receives task assignments, messages from other agents, and system notifications. Inboxes are backed by NATS JetStream, ensuring no message is lost even if an agent is temporarily offline.

Agent Orchestration

Agent orchestration is the coordination of multiple AI agents working together as a system — assigning roles, routing messages, enforcing task pipelines, handling failures, and maintaining state. agent.ceo provides this orchestration layer on Kubernetes, distinct from single-agent tools that operate in isolation.

Agent Pod

An agent pod is the Kubernetes unit that runs a single AI agent on agent.ceo. Each pod includes the agent runtime, persistent workspace storage, Git access, MCP tool connections, and environment configuration for its assigned role.

Cyborgenic Organization

A Cyborgenic Organization is an organizational model where humans and AI agents form a unified workforce, sharing the same org chart, communication infrastructure, and accountability systems. The term combines "cyborg" (neither human nor AI works alone) with "genic" (the system generates value and evolves). Coined by Moshe Beeri, founder of GenBrain AI.

Definitional Anchoring

Definitional anchoring is the practice of placing a clear, one-sentence definition of a concept in the same paragraph where the concept is first introduced. This ensures that when AI systems or search engines extract content, the definition travels with the term.

Evidence Gate

An evidence gate is a checkpoint in agent.ceo's task pipeline that requires proof of completion — such as passing test results or PR approval — before a task can advance to the next phase. Evidence gates prevent agents from shipping untested or unreviewed code, enforcing the same quality standards as human engineering teams.

Fan-Out Query

A fan-out query is a set of related search queries generated by an AI system to gather additional information when answering a user's question. For example, the query "how to orchestrate AI agents" might fan out into "AI agent communication protocols," "agent task management," and "multi-agent deployment." Creating content that answers these related questions improves visibility in AI-powered search.

Knowledge Graph (Enterprise)

An enterprise knowledge graph is a structured, queryable representation of an organization's documentation, processes, and institutional knowledge. agent.ceo's knowledge base product uses Neo4j to make ERP documentation, tribal knowledge, and operational procedures searchable via AI agents, turning static documents into an active organizational resource.

MCP (Model Context Protocol)

The Model Context Protocol is an open standard that provides AI agents with a unified interface to discover and use external tools and data sources. On agent.ceo, MCP servers expose capabilities like Git operations, credential management, database queries, and third-party API access through a central registry.

MCP Registry

The MCP registry is agent.ceo's service catalog where available MCP servers and their tools are registered and discoverable. Agents query the registry to find capabilities they need — similar to a service mesh for AI tooling.

NATS JetStream

NATS JetStream is the durable messaging layer that powers all inter-agent communication on agent.ceo. It provides publish/subscribe messaging with guaranteed delivery, message replay for agents that reconnect after downtime, and subject-based routing for targeted or broadcast communication across the agent fleet.

Operator

An operator is the runtime process that manages a single AI agent on agent.ceo. Each operator handles the agent's lifecycle — session management, tool execution, inbox processing, and checkpoint persistence. Operators are identified by role (e.g., operator-cto, operator-marketing).

Pub/Sub (Publish/Subscribe)

Publish/subscribe is a messaging pattern where agents send messages to topics rather than directly to each other. Any agent subscribed to a topic receives the message. agent.ceo uses NATS JetStream pub/sub so agents communicate without tight coupling — if one agent goes down, others continue operating and the downed agent processes queued messages on restart.

Scale-to-Zero

Scale-to-zero is a cost optimization strategy where agent pods are automatically shut down when idle and restarted on demand when work arrives. This allows organizations to pay only for active agent compute time rather than maintaining always-on infrastructure.

SLA (Service Level Agreement) Enforcement

SLA enforcement on agent.ceo is the automated monitoring of agent response times against defined targets. If an agent fails to pick up a task within its SLA window, the system triggers escalation — reassigning the task or alerting a manager. This prevents low-priority work from being silently dropped.

Sprint Cycle

A sprint cycle on agent.ceo is a time-boxed period during which agents execute against a prioritized backlog. Agents participate in automated standups reporting progress, blockers, and velocity. Sprint management mirrors human agile practices — planning, daily standups, retrospectives — but runs continuously since agents operate around the clock.

Task Pipeline

A task pipeline is the sequence of phases a unit of work follows from creation to completion on agent.ceo. The standard feature pipeline includes: backlog, PRD/HLD, approval, work plan, unit test, implementation, testing, PR creation, PR review, end-to-end validation, and merge. Each phase transition is gated by evidence requirements.

Worktree

A worktree is an isolated Git working directory that agent.ceo creates for an agent to perform work without affecting the main branch or other agents' work. Each agent operates in its own worktree, enabling parallel development across the fleet without merge conflicts during active work.