Most "AI agent + email" tutorials start the same way: connect the agent to a human's inbox over OAuth, hope the token doesn't expire mid-run, and pray the agent never replies to the wrong thread on someone's behalf.
There's a different model: give the agent its own email address. Nylas recently shipped Agent Accounts (currently in beta) — fully functional, Nylas-hosted mailboxes you create and control entirely through the API. Each one is a real [email protected] address that sends, receives, hosts calendar events, and RSVPs to invitations. To anyone interacting with it, it's indistinguishable from a human-operated account.
I work on the docs at Nylas, so I've spent a lot of time with this API. Here's a tour of what it does and how to get a mailbox running in a few minutes.
Why not just connect the agent to a human inbox?
You can — that's what OAuth grants are for, and they're the right tool when the agent works on behalf of a person. But a lot of agent workflows want a first-class identity instead:
System mailboxes (sales@,support@,scheduling@) that your app owns end-to-end. No OAuth consent screen, no user offboarding breaking your integration.
Ephemeral inboxes for test automation — provision a fresh address per run, sign up for a service, grab the OTP from the verification email, tear it down.
Per-customer identities in multi-tenant apps:[email protected],[email protected], each with its own send quota and sender reputation, all in one Nylas application.
A scheduling bot with its own calendar that proposes slots, sends invites, and shows up as a normal participant in Google Calendar, Microsoft 365, and Apple Calendar.
The key design decision: an Agent Account is just another grant. It gets a grant_id that works with every existing Nylas endpoint — Messages, Drafts, Threads, Folders, Attachments, Calendars, Events, Webhooks. If you've already built against connected accounts, nothing new to learn.
Create a mailbox with one API call
Every Agent Account lives on a domain — either a Nylas-provided *.nylas.email trial subdomain (instant, good for testing) or your own domain with MX and TXT records configured. With a domain registered, creation is a single request to the same Bring Your Own Auth endpoint Nylas uses for other providers, with "provider": "nylas". No refresh token needed:
curl --request POST \
--url "https://api.us.nylas.com/v3/connect/custom" \
--header "Authorization: Bearer $NYLAS_API_KEY" \
--header "Content-Type: application/json" \
--data '{
"provider": "nylas",
"settings": {
"email": "[email protected]"
}
}'
The response contains a grant_id — save it, it's the handle for everything else. The mailbox is live immediately with six system folders (inbox, sent, drafts, trash, junk, archive) and a primary calendar.
If you prefer a CLI, it's one command after nylas init:
nylas agent account create test@your-application.nylas.email
Receive email like any other grant
Inbound mail fires the standard message.created webhook — identical in shape to the same event for a Gmail or Outlook grant. Register one and Nylas calls your URL the moment a message arrives:
curl --request POST \
--url "https://api.us.nylas.com/v3/webhooks" \
--header "Authorization: Bearer $NYLAS_API_KEY" \
--header "Content-Type: application/json" \
--data '{
"trigger_types": ["message.created"],
"callback_url": "https://yourapp.example.com/webhooks/nylas"
}'
If your app handles both connected grants and Agent Accounts, branch on the grant's provider field ("nylas") to tell the deliveries apart. Polling GET /v3/grants/{grant_id}/messages works too if you don't want webhook infrastructure yet.
Send from the agent's own address
Outbound mail uses the same send endpoint as any connected grant:
curl --request POST \
--url "https://api.us.nylas.com/v3/grants/$GRANT_ID/messages/send" \
--header "Authorization: Bearer $NYLAS_API_KEY" \
--header "Content-Type: application/json" \
--data '{
"subject": "Hello from my Agent Account",
"body": "This message was sent by a Nylas Agent Account.",
"to": [{ "email": "[email protected]", "name": "You" }]
}'
The recipient sees a normal message from the agent's address — no "sent via" branding, no relay footer. Replies land back in the agent's inbox and thread normally, so multi-turn conversations (the thing LLM agents are actually good at) work out of the box.
The calendar is real too
Every Agent Account ships with a primary calendar that speaks standard iCalendar/ICS. The agent can:
- create events and invite people (
POST /v3/grants/{grant_id}/events) - accept or decline invitations it receives (
POST /v3/grants/{grant_id}/events/{id}/send-rsvp)
Because it's plain ICS under the hood, every major calendar client treats the agent as a regular participant. A scheduling agent can own its availability instead of borrowing a human's.
Guardrails: policies, rules, and lists
Letting an autonomous agent send email is exactly the kind of thing that deserves guardrails, and this is where Agent Accounts get interesting. You can attach policies that bundle send limits, spam detection, attachment restrictions, and retention settings — and assign one policy to many accounts. Rules match on sender, recipient, or message type and run actions like block, mark_as_spam, or assign_to_folder, with allow/block lists of domains, TLDs, or addresses referenced through an in_list operator.
So your support triage agent can block known spam domains at the SMTP stage before your LLM ever sees the message — which also keeps prompt-injection-laden junk out of the model's context.
Limits worth knowing (beta, free plan)
Send rate: 200 messages per account per day on the free plan (paid plans have no daily cap by default)
Storage: 3 GB per organization on the free plan
Retention: 30 days inbox, 7 days spam on the free plan (configurable via policy)
Outbound message size: 40 MB total
Domains: unlimited — one application can manage accounts across any number of registered domains
Agent Accounts are in beta, so the API may change before general availability.
Try it
The quickstart goes from zero to a sending-and-receiving mailbox in under 5 minutes. If you want recipes, the cookbook covers handling replies, multi-turn conversations, OTP extraction, and signing up for services autonomously.
Curious what people are building with agent-owned identities — if you've given an agent its own inbox (with Nylas or anything else), I'd love to hear how it went in the comments.
SOCIAL SHARE CARD GENERATOR