Frequently asked questions

OTP (Organization Transport Protocol) is a platform where organizations map their humans and AI agents in one accountability graph, author SOPs the agents inherit, and publish a structured Organizational Operating System their network can compare against. OTP sits at the organizational layer of the AI work-stack, above MCP (agent-to-tool) and A2A (agent-to-agent), and unlike those two it covers humans as first-class entities. Learn more in our founding essay.

Coordination intelligence is the collective, structured knowledge of how humans and AI agents within and across organizations should coordinate. It is captured in operational rules, documented failure modes, evidence-backed patterns, and the SOPs each role owns. It is to hybrid human-and-AI organizations what institutional knowledge is to traditional ones, except it is machine-readable, comparable, and inheritable. Read the full explanation.

CrewAI, AutoGen, and LangGraph are agent orchestration frameworks. They handle the plumbing of how agents execute tasks and pass messages. OTP operates at a different layer entirely. OTP captures the organizational intelligence that tells agents what to do, who owns what, and what happens when rules are broken. Think of it this way: CrewAI is how agents run. OTP is how agents know what the rules are. See the three-layer stack.

MCP (Model Context Protocol) connects agents to tools. A2A (Agent-to-Agent) connects agents to each other. OTP connects whole organizations, humans, AI agents, and the SOPs they share, to a network where accountability is queryable across orgs. MCP and A2A move bytes between agents; OTP moves accountability across organizations. Complementary, not competitive.

The Token Efficiency Ratio measures whether an operational rule is worth the tokens it consumes. Every claim loaded into an agent's context costs tokens. But a good rule prevents wasted cycles downstream. The ratio is: tokens saved by having the rule divided by tokens the rule costs to load. A ratio above 1.0 means the rule pays for itself. Below 1.0 means it should be cut or compressed.

Sign up, choose a template (Agent Army, Value Chain, or Org Chart), generate your OOS using the prompt on our guide page, and paste it into the publish form. The platform validates the format, extracts claims, scores quality, and publishes it to the Intelligence Graph.

Three templates: Agent Army for organizations with multiple specialized AI agents, Value Chain for process-oriented organizations, and Org Chart for traditional hierarchies integrating AI. See the glossary for details.

Confidence levels (HIGH/MEDIUM/LOW) express how certain you are about a claim. Evidence types describe how the claim was validated: measured results, repeated observation, human-defined rules, inference, or speculation. Together they make knowledge quality explicit and comparable.

A permanent badge carried by the earliest organizations to publish on OTP. The programme is closed, so the badge can no longer be earned. Publishing is still open to everyone and still puts your claims on the network.

From the browse page, select any two published OOS files to compare. The diff engine shows what is unique to each organization, what is similar, and where they conflict. This helps you learn from how others coordinate their AI teams.

The Intelligence Graph is a network visualization showing how coordination patterns connect across organizations. When two organizations share similar claims, those claims are linked. The graph reveals shared operational truths and unique approaches that no single organization could discover alone.

An 8-level framework (L1 through L8) measuring how sophisticated an organization's AI agent coordination is, from basic tab completion to fully autonomous agent teams. Based on the framework by Bassim Eledath. OTP displays each publisher's agentic level alongside their OOS. Read more in our blog post.

Any organization running AI agents. Solo operators with a single agent, small teams with 3-5 agents, and enterprises with dozens. The OOS format scales. If you have AI agents that need to coordinate, OTP captures how they should work together.

Published OOS files are visible to the community. You control what you publish. Use pseudonyms for your organization name if you prefer. Do not include proprietary information, trade secrets, or PII in your OOS. The platform includes a PII scanner that flags potential issues before publishing.

Yes. OTP provides a REST API and an MCP server for programmatic access to published OOS data, claims, search, and the Intelligence Graph. Publishers can generate API keys from their dashboard.