# Create a digest
> Answer two or three questions, edit the brief the wizard writes from them, and let the agent build the rules and the first report.
Create digests from the [Smart Digest](https://firehose.com/digests) page in the dashboard, under
**Digest** in the sidebar. Choose **New digest** to open the wizard.
Each digest uses one of your plan's active taps, so at the limit the button is unavailable and says so.
Pause a tap or upgrade the [plan](/billing/plans) to add a digest: a paused tap keeps its rules and
holds no slot.
## Pick a use case
The wizard opens on four starting points:
| Use case | What it tracks |
| --- | --- |
| **Track competitors** | Moves a competitor makes that you may need to respond to |
| **Monitor a topic** | Developments in a field, from the angle you care about |
| **Track your brand** | Where a brand or product is mentioned, and in what tone |
| **Something else** | Anything you describe in your own words |
**Start from scratch** skips the wizard and drops you straight into chat with the agent.
## Answer the questions
Each use case asks two or three short questions. Answer them in order:
- **Track competitors** — your company or product, which competitors to track, and which moves
matter (product and feature launches, pricing and packaging, positioning and messaging,
partnerships and acquisitions, leadership and hiring).
- **Monitor a topic** — the topic, which organizations to watch, and the angle that matters most
(technical developments, business and market impact, regulatory and policy, or all angles).
- **Track your brand** — the brand, its products or sub-brands, and where to look (news articles,
blog posts, review sites, forums and communities).
## Review the brief
The last step is **Review the brief**: the plain-prose brief the wizard composed from your answers,
in an editable box. It sets out the goal, the point of view, the subjects, what to keep and what to
drop, and what each item in the report records. Read it and fix anything that reads wrong — this is
the text the agent works from.
Name the digest, pick how often it refreshes and at what time, and choose **Create digest**.
Editing the brief here is faster than correcting the agent later. The Keep and Drop lines are the
starting point for the judging criteria, so a sharper bar in the brief means a shorter first
report.
## What the agent does next
The digest appears in the list and the agent starts work in the **Chat** tab. It writes as it goes,
in short concrete messages: what it found in the live data, which rule it proposes, what the
criteria say. In order, it:
1. Explores real matches to see what a rule for this topic would actually catch.
2. Creates and validates rules on the digest's own tap.
3. Dry-runs the digest against what it sampled, and adjusts the criteria.
4. Publishes the first report.
The report pane reads **No report yet** until that publish lands, then fills with the report. The
digest is listed as paused until then: its tap is live so the agent can test rules against real
matches, and the schedule starts with the first report.
## What success looks like
The digest's report pane shows a board of items with the fields your brief asked for, plus coverage
charts. The header reads **Updated** with a timestamp, and the schedule shows the cadence with the
next run. **Run now** and **Share** appear once the first report is published.
If the first report keeps too much or too little,
[refine it in chat](/smart-digest/refine-a-digest) — the digest stays, along with its rules and its
history.
## Next steps
Read a report, change its cadence, share it, and get it by email.
Change what's kept, the point of view, and the layout in chat.