Blog

Notes from Pidgeon

How-tos, the approaches compared, and what building an identity and an inbox for AI agents has taught us.

Everything we wrote

6 November 2026Gmail MCP server vs an inbox of the agent’s ownShould your AI agent use a Gmail MCP server, or an email inbox of its own?
4 November 2026What is an agent ID?What should an AI agent’s ID prove to the app or person it reaches, and why can’t they see who is accountable for it today?
4 November 2026Is OAuth enough to verify an AI agent?An access token says what a caller may do, but what does it say about which agent is calling, and who stands behind it?
4 November 2026Agent authentication vs user authenticationWhat changes when software, not a person, signs in or sends, and what should an app check before it lets an agent in?
4 November 2026Do you need an agent identity provider?If your company already runs an identity provider, what does an AI agent need that it does not already give you?
4 November 2026Centralised vs decentralised agent identityWho issues an AI agent’s ID, who checks it and who can take it back, in a central directory or on the agent’s own domain?
4 November 2026How to verify an email from an AI agentHow do you check whether an AI agent sent an email, who it acts for, and whether anyone changed it on the way?
4 November 2026How to verify which AI agent made a web requestWhen an AI agent calls your site or API, how do you tell which agent it is, who runs it, and that the request arrived unchanged?
4 November 2026Agent identity for agentic commerceWhen an AI agent buys something, what does the merchant need to know about who it is, and who decides whether it may pay?
4 November 2026A2A Agent Card identityAn A2A Agent Card says what an agent can do; what proves who runs it and who it acts for?
4 November 2026Enterprise-ready AI agents: what a security review asksWhat does a security review ask before AI agents email and act outside the company, and which controls hold if the model is fooled?
4 November 2026Email API vs inbox API for AI agentsShould your AI agent use a send-first email API with an inbound webhook, or an inbox API where it owns an address?
4 November 2026Inbox per agent vs a catch-all addressShould each AI agent or customer get its own inbox, or can one catch-all address and a database row do the job?
4 November 2026Multi-tenant email for AI agent platformsHow does an AI agent platform give every client its own inboxes, domain, keys and sending reputation without one client hurting another?
4 November 2026Why your AI agent's emails go to spamWhy do emails sent by an AI agent land in spam, and what should you check first to fix it?
4 November 2026Building agent email on Amazon SESWhat do you have to build yourself to run AI agent inboxes on Amazon SES, and when is building it the right call?
4 November 2026How to trigger an AI agent when an email arrivesHow does an AI agent start work the moment an email lands, with or without a public URL?
4 November 2026Idempotent email sending for AI agentsWhy do AI agents send the same email twice, and how do you make every send happen exactly once?
4 November 2026Email threading for AI agentsHow does an AI agent keep an email conversation in one thread, and what happens when a person steps in?
4 November 2026MCP vs SDK vs API for AI agent emailShould an AI agent get email through an MCP server, an SDK, the REST API or a CLI, and what does each cost?
4 November 2026How to give your AI agent its own calendarShould an AI agent schedule from your calendar, or have its own, and how does an agent's calendar work?
30 October 2026What the EU AI Act asks of email agents by 2 DecemberArticle 50 has applied since 2 August 2026: an agent that writes to people must say it is an AI, and whom it acts for, from the first message. Its text must also carry a machine-readable mark, due by 2 December 2026 for systems already on the market.
29 October 2026How to give your AI agent its own email addressGive the agent an inbox of its own, not a login to yours: create an address by API, receive mail as a webhook or a call that waits, and reply from that address in the same thread.
29 October 2026Protecting an AI agent from prompt injection in emailTreat every email as untrusted input and defend in three layers: screen it before the model reads it, hand the model the visible text, and limit what an instruction that gets through can do.
29 October 2026Is it safe to let an AI agent read your email?It can be, if you limit what the agent can read and what it can do with it. The risk is not the model seeing mail; it is that anyone who can email you can now give instructions to software acting for you.
29 October 2026Email MCP servers for sending and receiving, comparedPick by whose inbox it is: an agent inbox server such as Pidgeon’s or AgentMail’s for an address the agent owns, a Gmail server to work inside yours. Then check five things before you connect one.
29 October 2026How to give Claude Code and Cursor an email addressAdd an email MCP server and give the coding agent an address of its own. In Claude Code it is one command, in Cursor one entry in mcp.json, and a sign-in where you choose what it may do.
26 October 2026How a browser agent receives an email verification codeSign up with an address the agent owns, make one call that waits for the email, and read the code already pulled out of it: no polling loop, no regex, and no person fetching codes from their own inbox.
26 October 2026Human-in-the-loop email agentsAn approval queue checks the agent’s replies. A hand-off gives the conversation to a person. Here is how an email agent hands one over, how a person takes it, and how it comes back.
26 October 2026An email sandbox for AI agentsTest mode in Pidgeon is the same API with a different key: the whole pipeline runs, nothing leaves, nothing counts, and you can cause a bounce, a refusal or a lost answer on purpose.
26 October 2026An EU email API for AI agentsAn agent’s inbox is personal data from its first message. Pidgeon stores it in the EU on every plan, is run by an EU company, and applies its DPA automatically.
25 October 2026Email for OpenClaw that keeps workingAn email to your agent’s Pidgeon address starts an OpenClaw conversation, one per email thread. The agent’s answer goes back as a reply in that thread, and a reply to the reply carries on the same conversation.
16 October 2026The best email APIs for AI agents in 2026An agent that does real work needs email. Which API depends on one question first: does the agent need its own inbox, somebody else’s, or only a way to send?

Give your agents an inbox

Every inbox comes with its ID, and both are free to start.

Create your first inbox, it’s free