# Content pipeline

Use Rowset when agents need to track content briefs, drafts, review state,
canonical URLs, and publishing evidence without forcing the workflow into a CMS.

## Starter shape

Create a `content_queue` dataset indexed by `slug`.

| slug | content_type | stage | owner | target_keyword | canonical_url | publish_date | notes |
| --- | --- | --- | --- | --- | --- | --- | --- |
| mcp-dataset-api | blog | review | Scribe | MCP dataset API |  |  | Needs examples |
| agent-crm-guide | use case | draft | Scribe | agent CRM | /use-cases/personal-crm |  | Outline approved |
| feedback-board | landing | idea | Beacon | feedback board |  |  | Use-case angle |

## Agent jobs

- Create briefs from research and customer notes.
- Move items through review and publish stages.
- Attach canonical URLs, owners, and completion evidence.
- Export the queue for editors, scripts, or downstream systems.

## Workflow rules

Define stages before agents start editing: `idea`, `brief`, `draft`, `review`,
`scheduled`, and `published` are enough for most small teams. Add instructions
for when agents can create drafts, when they must wait for approval, and where
published URLs should be recorded.
Use the guide to [structuring dataset instructions for AI
agents](/blog/structure-dataset-instructions-ai-agents) when those stage rules
need to survive across agent sessions.

## Connect it

Use [MCP access](/docs/connect-mcp) for agent planning and updates. Use the
[Dataset API](/docs/dataset-api) when a publishing script needs a structured
queue.