Email you can test against people, without mailing anyone

Why it exists

A lot of software lives or dies by email: sign-up and password-reset flows, receipts, sales sequences, help desks, CRMs, and now AI agents that read and write mail on someone's behalf. Testing that software is awkward. A capture inbox shows you what your app sent, but nobody answers. Real inboxes answer, but then you are mailing real people, or juggling a pile of test accounts, and every mistake goes out for real.

simmail sits in between. It accepts the mail your app sends and keeps it, like a capture inbox, and it also puts people behind the addresses. They have habits, moods and memories of earlier threads. They take their time to reply, follow up on promises, and sometimes do the inconvenient thing. Around them is a world that writes in on its own: leads, support requests, spam and phishing. You can add faults such as greylisting, rate limits and bounces, and move the clock forward to see what happens next week.

All of it stays inside simmail. The mail server has no relay and no outbound SMTP client, and the build fails if one is ever linked in. Mail addressed to any domain is kept in your universe and goes nowhere else.

Who makes it

simmail is made by Plexus Labs, a small software team. The same people write the mail server, the simulation, the dashboard and these docs, and read what you send to support.

Principles

It never sends real mail

This is a design rule, not a setting. There is no code path that delivers mail to a real mailbox, and the test suite checks that none is added. The connections simmail does make outward are the AI calls that write replies, link clicks by simulated people (which can be limited by rules), and webhooks to URLs you set up. None of them carry mail.

Your data is yours

The mail in a universe is your test data. You can read every message over IMAP or the API, set how long it is kept within your plan, and delete a universe with everything in it. We do not use it for anything other than running your simulation.

Limits are written down

Every plan's limits are listed on the pricing page and explained in the docs, including what happens when you reach one. When AI actions run out, people keep replying from templates, and the dashboard says so. When the AI fails, the fallback is visible rather than hidden.

Get in touch

Questions, feedback or a bug: contact us. Security reports go to [email protected]; see Security.