Skip to main content

Documentation Index

Fetch the complete documentation index at: https://datost.com/docs/llms.txt

Use this file to discover all available pages before exploring further.

Slack is Datost’s primary surface. You ask questions the same way you’d ask a teammate — with an @mention in a channel, a DM, or a slash command — and Datost replies in-thread with the answer, the SQL it ran, and a link to the full response in the web app.
If you haven’t installed the Slack app yet, start at Install Datost in Slack. Everything below assumes the app is installed and at least one data source is connected.

​
The four ways people use Datost

Mentions in channels

@Datost what was revenue last week? in any channel where the bot is a member. Replies land in-thread. Follow-ups in the same thread reuse context.

Direct messages

DM the Datost app for a private conversation. No @mention needed — every message is treated as a question.

Slash commands

/datost link to connect your account, /datost schedule to set up recurring reports, /datost help for the full list.

Scheduled reports

Recurring prompts that run on a cron and deliver results to a channel, DM, or group chat. Created via /datost schedule.

​
The assistant thread (native Slack AI surface)

If your workspace has Slack’s native AI assistant UX enabled, Datost shows up in the right-hand assistant pane. Opening it starts an assistant_thread — a dedicated, scoped conversation that’s aware of whatever channel you had open when you launched it. Every message you send in that pane is treated as a question; Datost streams a typing status, posts the answer, and can suggest follow-ups. Changing the surrounding channel sends a context_changed signal so the next question can pick up the new channel’s context automatically. Mentions in regular channels use the standard thread model: Datost replies as a threaded message under your question, and any reply you add to that thread is treated as a follow-up.

​
Where results go

  • Thread replies — the primary answer lands as a Slack message in the same thread as your question, with inline charts (bar, line, area, pie) rendered as Block Kit images.
  • “View full response” link — every reply includes a deep link to the hosted response page in the web app, where you can see the full tool trace, SQL, sortable result tables, and timing. See Viewing Slack responses.
  • Follow-ups — reply in the same thread. Datost reuses the prior query’s sandbox when it’s still warm, so iterations stay fast.
  • Cancel a running query — react to Datost’s “thinking” message with ❌ and it will stop.

​
Who can ask Datost

By default, any member of the Slack workspace can @mention Datost in a channel the bot is in, or DM it. There’s no per-user allowlist at the Slack layer. Identity and attribution, though, depend on account linking:
  • Linked users (ran /datost link and completed the web-app flow) are attributed to their Datost account. Their queries show up in the web app under their user, and they can create and own scheduled reports.
  • Unlinked users can still ask questions — Datost resolves the workspace’s Datost org from the Slack team ID and answers using the org’s connected data sources. The query is recorded against the Slack user, not a Datost user.
Org admins can restrict Datost to specific channels by only inviting the bot to those channels. Permissions on the underlying data (which warehouse, which tables) are enforced in the web app per context, not per Slack user.
Rate limits are enforced per Slack user and per channel (hourly and daily). If you hit one, Datost replies with a retry-after hint.

​
Next

Mentions & threads

How thread context, follow-ups, and the assistant pane behave.

Slash commands

Every /datost subcommand, with examples.

Scheduled reports

Recurring research delivered to Slack on a cron.

Install Datost in Slack

OAuth, scopes, and adding the bot to channels.