Assistants and bots
Applies to: SimpleTwo 0.9.x · Checked: 20.09.2026
An agent is a program that works in the conversation instead of you or beside you: it triages requests, answers routine questions, brings notifications from other systems. Everything created in your name is collected on the My agents screen.
1. Two kinds, and the difference matters
| Kind | What it is | What bounds it |
|---|---|---|
| Acts in your name (assistant) | a session of your account: it reads everything you can read and writes as you | nothing but your own rights: it is a member of nothing, its reach is your reach |
| Acts under its own name (worker) | a separate participant | it sees only the chats it was added to, like any other member |
The first row is why this screen exists: an assistant is you. A "these chats only" limit on it is a contradiction, and the app says so rather than pretending the limit works.
2. What the card shows
- May — the permissions in plain words: read, write, react, create chats and change their membership — and in whose name.
- Reach — whose: yours, or its own.
- Expiry — when it runs out, or "no expiry".
- Last acted — or "never acted".
- Sends events to a webhook — when the agent is wired to an external system.
A permission this version of the app does not know is shown as-is rather than hidden: an unknown right is a reason to stop the agent, not a reason to hide it.
3. States, and what to do about them
| State | What it means | What to do |
|---|---|---|
| Working | all fine | — |
| Stopped | you stopped it: it neither writes nor acts | Start restores every permission unchanged |
| Expired | its token ran out | it needs a new one |
| Blocked | the organisation forbade agents acting in people's names | it cannot connect until that is lifted |
Stop is a reversible pause. Delete is not: the token stops working immediately and the permissions are removed. If you only want to pause it, stop it rather than deleting it.
The list and the stop button live in the app for exactly that reason: somebody who has to write to support to stop their own agent will not stop it. The same list appears on the Sessions screen beside your devices — "who can act as me" is answered in one place.
4. Marks on messages
For an assistant you can choose whether others see a mark on messages sent in your name.
Whichever you choose, every message records which agent created it, and that record cannot be switched off.
5. Secret chats are closed to agents
No agent — including your own — can read secret chats: they are encrypted for your devices and the server cannot decrypt them. That is a property of the encryption, not a setting that can be relaxed.
Next
- Chats, threads and topics
- Bots and automations — the same thing from the administrator's side