Compare
Pidgeon vs AgentMail
Both give an AI agent its own inbox through an API. Here is where they differ — on what happens when a person needs to step in, on what screens a message, and on price.
The short version
AgentMail and Pidgeon both give an AI agent an email address it owns, through an API. Either will receive mail, thread it, store it, open in a mail client over IMAP and SMTP, and tell your code when something arrives.
The difference is who else the inbox is for. AgentMail is built for agents that work alone: software creates inboxes and works them through REST, webhooks and WebSockets, and a person reviews drafts. Pidgeon is built for agents that work alongside people: every thread has an owner — a person, an agent or an automation — and handing a conversation over is one call, so the agent stops when a person takes it and picks up when they hand it back.
Choose AgentMail if you want the most established option, the longest list of framework integrations, or a SOC 2 report today. Choose Pidgeon if a human will ever need to read along, step in, or own part of the conversation — or if you want your own domain on the free plan.
At a glance
Prices and limits as published on 17 October 2026. AgentMail prices in US dollars, Pidgeon in euros.
| Pidgeon | AgentMail | |
|---|---|---|
| Free plan | 3 mailboxes, 1 custom domain, 3,000 messages a month | 3 inboxes, 3,000 emails a month; no custom domain on the pricing page |
| First paid plan | €20 a month: 10 mailboxes, 10 domains, 10,000 a month | $20 a month: 10 inboxes, 10 domains, 10,000 a month |
| Next plan | €200 a month: 150 mailboxes, 150 domains | $200 a month: 150 inboxes, 150 domains |
| API, SDKs, CLI | REST, TypeScript and Python SDKs, CLI | REST, TypeScript and Python SDKs, CLI |
| MCP server | Hosted (pidgeon.ai/api/mcp) and npm | Yes |
| Agent signs itself up | Yes — two calls and a code from a human | Yes — and a human claims the inbox |
| IMAP and SMTP for mail clients | Every plan, with an app password for one mailbox | Yes, with an API key as the password |
| Real-time events | Signed webhooks, event stream, WebSockets | Webhooks, WebSockets |
| Threads with an owner (human, agent, automation) | Yes, with assignment events | Drafts for review |
| Mail that tries to steer the agent | Firewall verdict on every inbound message | Agent Armor, beta by request |
| Per-address permissions | Send and receive rules; keys that can only mint narrower keys | Allow and block lists; scoped keys |
| Multi-tenant grouping | Pods, with a key per pod | Pods |
| Calendar | Every mailbox | Every inbox, beta in the US |
| "Sent via" footer on free mail | Yes | Yes |
Where Pidgeon is different
Hand-off is state, not a workaround
A Pidgeon thread has an owner — human, agent or automation — and changing it is one call. thread.assigned and thread.updated tell your agent the moment a person takes a conversation, so it stops replying; handing it back clears the name and the agent picks up where it left off. The hand-off guide walks through all five steps.
That matters the first time a customer replies with something the agent should not answer.
A person's mailbox and an agent's, on one domain
hello@ a person reads and receipts@ a program owns are the same kind of object in the same account, with one log and one bill. A person opens theirs in Apple Mail, Outlook or Superhuman with an app password scoped to that one mailbox — not an API key, which can do everything the key can do, saved in a mail client's settings.
Permissions that narrow and never widen
An API key can be scoped to one address and one level of access, and a key can create a narrower key but never a wider one. Revoking a key revokes everything it created. Send rules decide who an address may write to — "only to people at acme.com" is a rule, not code in your agent.
A verdict on every message, today
Anyone who can email your agent can try to instruct it. Pidgeon Firewall reads every inbound message first and names what it found — hidden text, instructions addressed to an AI, look-alike senders, requests for secrets — so your agent can check before it acts. The Firewall guide has the signals. AgentMail's equivalent, Agent Armor, is in beta by request.
Your domain, on the free plan
One custom domain is part of Free, with SPF, DKIM and DMARC written for you, and one click if your registrar supports Domain Connect. AgentMail's pricing page lists no custom domain on its free plan.
Receiving is never metered
Mail arriving at a Pidgeon address does not count towards anything. Sending is capped by plan and by a trust score that rises with use; receiving is free because charging for the thing that makes an inbox worth having would be the wrong business.
Where AgentMail is ahead
Worth knowing before you decide, because a comparison that only lists our strengths is not a comparison.
- Time in market. AgentMail launched in 2025, is backed by Y Combinator, and names over a dozen production customers. Pidgeon launched in October 2026.
- Integrations. AgentMail has plugins or guides for Claude Code, Codex, Cursor, OpenClaw, Replit, the Vercel Marketplace, LangChain, the OpenAI Agents SDK and Google ADK. Pidgeon has its SDKs, CLI, MCP server, agent skill and a Vercel AI SDK tool set; more framework packages are being built now.
- SOC 2 Type II. AgentMail has a report, available from its Startup plan up. Pidgeon does not have one yet.
- EU data residency. AgentMail offers an EU region on Enterprise.
- More around the inbox. A file drive per inbox, and AgentID, a sign-in for agents built on the address.
Which should you choose?
- An agent that only ever talks to other software, at scale, with no person in the loop: either works. The allowances are the same at each price; compare the integrations you need.
- An agent whose conversations a person has to read, review or take over: Pidgeon. Thread ownership and hand-off are what it is built around.
- A product that gives each of its own customers an address: both have pods. Pidgeon's key per pod, and keys that can only narrow, make it safe to hand a customer a key of their own.
- You need a SOC 2 report this quarter: AgentMail.
- You want to try it on your real domain before paying: Pidgeon's free plan includes it.
Moving from AgentMail
The concepts map one to one: an inbox is an address, a pod is a pod, a webhook is a webhook. Create the addresses through the API or let your agent do it, point your domain's MX records at Pidgeon, and keep AgentMail running until the last thread in flight has finished. The quickstart is about five minutes; the MCP guide covers Claude Code, Cursor and Codex.
Or paste one line into your coding agent and let it do the rest:
Read pidgeon.ai/skill.md and get yourself an email address.
We make Pidgeon, so weigh this accordingly. Every claim about AgentMail comes from its own pricing page, documentation and blog, checked on 17 October 2026. We re-check this page every month. If something here is out of date, write to support@pidgeon.ai and we will fix it.