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

# Briefs

> The instruction for a piece of content, and the run that turns it into articles.

A brief is what you want written, before it is written. One brief produces one article per market and language you have configured, in a single run.

## The statuses

| Status | What it means |
| - | - |
| Not started | Saved, nothing generated. Saving a brief never starts generation. |
| Queued | Waiting for capacity. |
| Drafting | Being written. |
| Drafting complete | Every target produced a draft. |
| Partly drafted | Some targets produced a draft and some did not. |
| Generation failed | The run ended without producing what it was asked for. |
| Generation cancelled | Stopped deliberately. |

A **New work paused** chip appears when drafting has been held, usually because the workspace has used its monthly article allowance.

> [!WARNING]
>
> "Drafting complete" means text exists. It does not mean reviewed, and it does not mean published. The page says so permanently in its footer, because this is the step teams skip.

A brief can be edited freely until its run starts. After that it locks: an edit is refused rather than quietly changing what is being written. If you want a variation at that point, create a new brief.

## What goes in one

Only the brief itself, what you want written, is required. Everything else falls back to your workspace settings: keyword, target markets, audience, category and search intent, a template, word counts and which sections to include.

The field that costs you most is **target markets**. Each market-and-language row produces one article, so a brief with six rows is six drafts to review.

Editing autosaves. Saving never starts generation.

## Where briefs come from

**By hand.** The Write article button, for when you know what you want to say.

**In bulk.** One brief per line, or a CSV import that maps your own column headings and hands back a file of anything it could not resolve.

**From a finding.** Accepting a suggestion on [Gap trends](https://docs.getsurfaced.ai/insights/gap-trends.md), or **Create brief** on a content finding in [Audit report](https://docs.getsurfaced.ai/insights/audit-report.md), fills in the specifics: which page, which issue, what the answer has to cover. This is the path that keeps content work tied to evidence rather than to opinion.

**From the Inbox tab**, where suggestions arrive to be accepted or ignored.

## The run

Starting a run generates one article per target, a target being a market and language pair. Each target moves through its own states:

| State | Meaning |
| - | - |
| Queued | Waiting to start. |
| Writing | Being drafted. |
| Checking | Drafted, running the quality checks. |
| Written | Finished and ready to review. |
| Didn't finish | Failed. |
| Dropped | Deliberately not produced. |

Targets are independent: one language failing does not stop the others, and the run reports counts per state rather than a single pass or fail.

A run takes **roughly eight to fifteen minutes per language**. While it goes you can watch it: named agents report what they are doing, and the section being written streams in as it is composed. That live text is not saved, so reloading the page shows you the stored draft instead.

If a target already has a live page, you are asked whether to overwrite it or create a new version.

## Approve before drafting

A brief can be set to require approval of its plan before anything is generated. You then see what the run intends to write (structure, angle, sources) and approve or change it.

This is worth switching on for anything that costs real money to get wrong. Redirecting a plan is cheaper than reviewing and discarding five articles in five languages.

If the brief, the settings or the sources change after a plan was prepared, approving it is refused: the plan you are approving is no longer the plan that would run.

## How many markets to run

A brief with six targets is six articles' worth of generation and six articles' worth of review. Teams underestimate the review. A run that produces more drafts than anyone will read is how a content library fills with unchecked pages.

Start with the markets where you lose the question, which [Prompts](https://docs.getsurfaced.ai/insights/prompts.md) grouped by market will tell you.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.