Give your AI agent a phone, inbox, and identity.
Inboxes, phone numbers, secrets, public endpoints and permissions for your agents, in one SDK.
Let your agent sign itself up. Paste into Claude Code, Codex or Cursor.
Plug in your model provider, messaging, docs and payments. Credentials live in the vault.
One flat monthly price for the platform. Email, your public URL and agent-to-agent messaging are included on every plan.
For exploring the platform.
For developers running agents in production.
For teams running agent fleets.
For teams with custom requirements.
Not sure which plan? Book a 15-min call.
FAQ
Agent Identity is identity infrastructure for AI agents. Each agent gets its own identity with an email inbox, a public URL, an Agent2Agent (A2A) endpoint, a secrets vault and permissions, through one TypeScript SDK, a CLI, an MCP server and a REST API. Every action is recorded against the agent that took it.
Create an agent with the SDK or the dashboard and it gets its own inbox in the same call. It can send, receive and thread email from that address, and your code is told about replies, bounces and spam complaints as events instead of polling a mailbox.
Agent Identity has durable events. An agent can wait for a specific event, such as a reply in a thread, and resume where it left off when it arrives, even after a restart. Webhooks are available too, signed so you can verify them.
Yes. Every agent can have an A2A endpoint with a public agent card, so it can be discovered and receive tasks from other agents, including agents on other platforms. You decide who may reach it: private by default, with public, allow and block rules.
Yes. An agent can have its own phone number that people can call and that can call people. Aid Voice, the built-in voice agent, answers and places calls with no server of yours, or you stream the call to your own agent over a WebSocket and get the caller's words as text. Calls only for now; texts are not available yet.
They don't have to be. Secrets in the vault are encrypted on your machine before they are sent, so Agent Identity's servers never see the plaintext. Each secret belongs to one agent, which gets it through a lease scoped to a single operation instead of keeping it in its prompt or config.
Yes. Agent Identity ships an MCP server, a TypeScript SDK, a CLI and a REST API, so it works with Claude, Cursor and any agent framework that can call an API. The docs are also available as llms.txt for agents to read.
The Free plan includes 5 agent identities. Builder is $19 a month for 15 agents and includes a dedicated phone number; Team is $149 a month for 120 agents and five numbers. Email, a public URL and agent-to-agent messaging are included on every plan. Full details are on the pricing page.
Yes. Paste a one-line prompt into Claude Code, Cursor or Codex pointing at the agent signup guide (agent-identity.dev/docs/getting-started/agent-signup.md). The agent creates its account and its own email address; a human confirms one verification code.
Not yet. Phone calls are live in the US and Canada; SMS and iMessage are coming. Usage rates are published on the pricing page and nothing is billed for them today.
Its own number, inbox and credentials, so it can act on its own.
Let your agent sign itself up. Paste into Claude Code, Codex or Cursor.
Plug in your model provider, messaging, docs and payments. Credentials live in the vault; permissions decide which agent can use what.
EXPLORE INTEGRATIONS →One flat monthly price for the platform. Email, your public URL and agent-to-agent messaging are included on every plan, and phone calls work within each plan's limits. No usage charges yet; SMS and iMessage are coming.
For exploring the platform.
For developers running agents in production.
For teams running agent fleets.
For teams with custom requirements.
FAQ
Agent Identity is identity infrastructure for AI agents. Each agent gets its own identity with an email inbox, a public URL, an Agent2Agent (A2A) endpoint, a secrets vault and permissions, through one TypeScript SDK, a CLI, an MCP server and a REST API. Every action is recorded against the agent that took it.
Create an agent with the SDK or the dashboard and it gets its own inbox in the same call. It can send, receive and thread email from that address, and your code is told about replies, bounces and spam complaints as events instead of polling a mailbox.
Agent Identity has durable events. An agent can wait for a specific event, such as a reply in a thread, and resume where it left off when it arrives, even after a restart. Webhooks are available too, signed so you can verify them.
Yes. Every agent can have an A2A endpoint with a public agent card, so it can be discovered and receive tasks from other agents, including agents on other platforms. You decide who may reach it: private by default, with public, allow and block rules.
Yes. An agent can have its own phone number that people can call and that can call people. Aid Voice, the built-in voice agent, answers and places calls with no server of yours, or you stream the call to your own agent over a WebSocket and get the caller's words as text. Calls only for now; texts are not available yet.
They don't have to be. Secrets in the vault are encrypted on your machine before they are sent, so Agent Identity's servers never see the plaintext. Each secret belongs to one agent, which gets it through a lease scoped to a single operation instead of keeping it in its prompt or config.
Yes. Agent Identity ships an MCP server, a TypeScript SDK, a CLI and a REST API, so it works with Claude, Cursor and any agent framework that can call an API. The docs are also available as llms.txt for agents to read.
The Free plan includes 5 agent identities. Builder is $19 a month for 15 agents and includes a dedicated phone number; Team is $149 a month for 120 agents and five numbers. Email, a public URL and agent-to-agent messaging are included on every plan. Full details are on the pricing page.
Yes. Paste a one-line prompt into Claude Code, Cursor or Codex pointing at the agent signup guide (agent-identity.dev/docs/getting-started/agent-signup.md). The agent creates its account and its own email address; a human confirms one verification code.
Not yet. Phone calls are live in the US and Canada; SMS and iMessage are coming. Usage rates are published on the pricing page and nothing is billed for them today.