> ## Documentation Index
> Fetch the complete documentation index at: https://docs.lettr.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Activity Log & Undo

> Every change Adamko makes is recorded in a team-wide activity log, and most changes can be undone with one click — as long as nothing has touched them since

An assistant that creates real things in your account needs a paper trail. Adamko has one: every entity he creates, updates, or deletes is recorded in an activity log visible to your whole team, and most changes carry an **Undo** button.

## The Activity Log

Open it from the **Activity** button in the drawer header. Entries are grouped by the request that caused them — you see who asked, when, and the exact prompt they typed, followed by what Adamko did:

> **Erik · 2h ago** — *"set everything up for me"*
> Created 4 templates · Created API key "Adamko API key" · Created domain mail.example.com

The log covers seven kinds of entities: campaigns, templates, brand kits, sending domains, API keys, contact properties, and topics. It's **team-wide** — everyone on the team sees what Adamko did for anyone, which matters because any team member can use him. Chat conversations themselves stay private per user; only the changes are shared.

The log is permanent. Undoing a change marks the entry as **Undone** (struck through) rather than erasing it — the history of what happened, including reversals, is always intact.

## Undo

Each loggable change shows an **Undo** button while undoing is still safe. There's no time limit — instead, the rule is **"untouched since"**: a change can be undone as long as nothing meaningful happened to the entity after Adamko made it.

| Change                         | Undoable while                           | Undo does                                             |
| ------------------------------ | ---------------------------------------- | ----------------------------------------------------- |
| Campaign drafted               | Still a draft and never edited by anyone | Deletes the draft                                     |
| Template created               | Never edited (no version changes)        | Deletes the template                                  |
| Brand kit generated or updated | Kit still exists                         | Deletes it, or restores the previous version          |
| Sending domain created         | Not yet DNS-verified                     | Removes the domain from your account                  |
| API key created                | Key still exists                         | **Revokes the key** — anything using it stops sending |
| Contact property created       | Never edited                             | Deletes the property and its values                   |
| Topic created                  | No contacts subscribed yet               | Deletes the topic                                     |

The logic behind the exceptions: a verified domain means you adopted it; an edited template contains your work now; a topic subscription is a real consent record, not an AI artifact. Once a human has built on top of Adamko's change, one-click reversal would destroy human work — so the button disappears, and from then on you manage the entity normally.

Undo always asks for confirmation first, stating the consequence plainly ("This permanently revokes the API key — anything still using it will stop sending."). Eligibility is re-checked at the moment you confirm, so a teammate editing the template a second earlier turns the undo into a friendly refusal rather than a lost edit.

<Warning>
  Undoing an API key is a real revocation. If you've already deployed the key to production, undoing the creation will stop your application from sending immediately.
</Warning>

## What Doesn't Appear in the Log

Read-only work — answering questions, running diagnoses, searching docs — isn't logged as activity, because nothing changed. Separately from this user-facing log, every tool call Adamko makes (including read-only ones) is audited internally with its inputs and outputs, which is part of how the feature is monitored for correctness.

## Next Steps

<CardGroup cols={2}>
  <Card title="Privacy, permissions & limits" icon="shield-halved" href="/learn/ai-assistant/privacy-and-limits">
    What data Adamko sees and who can use him
  </Card>

  <Card title="What Adamko can create" icon="paintbrush" href="/learn/ai-assistant/content-creation">
    Templates, campaigns, and the brand kit
  </Card>
</CardGroup>
