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.
When an agent gathers research from recurring APIs, feeds, files, or approved web sources, use a separate AI data collection control plane for source authorization, capture checkpoints, provenance, and acceptance. Put only accepted findings into the content queue so research acquisition and editorial state do not become one ambiguous workflow.
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 when those stage rules
need to survive across agent sessions.
Connect it
Use MCP access for agent planning and updates. Use the Dataset API when a publishing script needs a structured queue.