Plans and limits

Your organization’s plan sets limits (how much) and features (what is included). Every universe in the organization shares them.

Limits

Limit	Counts	Resets
AI actions	Each time the AI writes or decides something: a reply, a person it invents (a stranger or a World sender, in the same call as their first email), a case story, an inbound email the World writes, reading an attachment or opened page (one per 10 pages, read once per universe however many people get it), and settling a close call such as whether someone opens a link. What the rules decide on their own is free.	Monthly
…with the better model	Of the AI actions, replies written with the stronger model, by people set to AI model: better. When they run out, those people use the standard model.	Monthly
Learned from mail	Personas learned from an address’s mail.	Monthly
Mail received	Messages your apps send into simmail, over SMTP or with `POST /api/send`.	Monthly
Universes	Universes in the organization.
People	Personas across all your universes, including strangers and World senders.
App logins	SMTP and IMAP logins across all your universes.
Members	People in the organization.
Retention	The longest that mail and activity are kept, in days.
Inbound streams	World streams that can be on at once.
Inbound emails a day	Emails the World sends your universes each day (UTC).	Daily

Monthly allowances reset at the start of each calendar month, UTC.

Extra AI actions

Owners and admins can buy extra AI actions on Plan & billing: 5,000 for $15. They never expire, and are used only after the plan’s monthly allowance runs out.

What doesn’t use AI actions

Most of a person’s small decisions are made by rules: whether they open a link, whether they look you up, whether an email is automated. Only the close calls go to the AI, and only on plans with AI decisions; on Free, the rules decide everything. People don’t answer automated email such as verification codes and receipts, so those cost nothing (they may still open its link). Reminders and out-of-office replies are written from templates.

What happens at a limit

Limit reached	What you see
AI actions	People keep replying, but from templates instead of the AI, until the month resets or you upgrade.
Mail received	SMTP refuses new mail with `452 4.5.3` and the plan’s message. `POST /api/send` answers with an error.
People, logins	Creating one more answers `400` with a message such as “the Starter plan includes 250 people; upgrade for more”.
Universes, members	Adding one more answers `402`.
Inbound streams	Saving the World with too many enabled streams answers `400`.
Inbound emails a day	The World stops sending until the next day (UTC).
Retention	Mail and activity older than the limit are removed, whatever the universe’s own setting.

Features

Feature	What it unlocks
AI decisions	The AI settles people’s close calls (opening a link, looking you up); without it, rules decide.
Lifelike behaviour	People click links, look up senders, make and keep promises, and nudge you. Changing link rules.
Junk consequences	Spam scores decide Inbox or Junk for each person. See Junk and spam.
Full inspection	The compatibility and accessibility details of the health report, and people reacting to them.
Chaos profiles	Provider scenarios, greylisting, rate limits, dropped connections and time windows in faults.
MCP	The MCP endpoint for AI agents.
Webhooks	Setting up webhooks.
Insights	Insights and shared reports.
Audit	The organization’s audit trail.

Calling something your plan does not include answers 402 with “this is not included in the … plan; upgrade to use it”.

When simmail launched, the plans included these features:

Feature	Starter	Pro	Team	Business
AI decisions	✓	✓	✓	✓
Lifelike behaviour		✓	✓	✓
Junk consequences		✓	✓	✓
Full inspection		✓	✓	✓
MCP		✓	✓	✓
Webhooks		✓	✓	✓
Chaos profiles			✓	✓
Insights			✓	✓
Audit			✓	✓

Check the pricing page for today’s plans.

Changing plans

See Billing.