Manual content research, scripting, and production removed, with a person approving each step.
Agents research topics, draft scripts, route them for approval, then render videos. A hands-free content loop with human checkpoints.
Sector
Marketing
Timeline
4 Weeks
Work area
Content and marketing operationsProject
AI Content Strategist Pipeline
The problem
Producing video content on a schedule meant the same manual cycle every week: finding topics, researching them, writing scripts, getting approval, and producing the video. The team spent most of its time on the repetitive steps rather than on the creative decisions.
A multi-agent content pipeline that researches trending topics, structures scripts, routes them through an approval workspace, and renders final videos. Human-in-the-loop checkpoints keep quality high while removing daily production bottlenecks.
What we built
- 01Research agentFinds and scores trending topics for the brand, with the sources behind each idea.
- 02Scripting agentTurns an approved topic into a structured script draft in the brand’s voice.
- 03Approval workspaceDrafts land in Notion for a person to edit and approve; nothing moves on without sign-off.
- 04Video renderingApproved scripts are rendered into videos automatically and queued for publishing.
Technology
- OpenClaw
- Notion
- HeyGen
- n8n
- GCP
Methods
- Agent Workflows
- Human-in-the-Loop
- Content Ops
How it works
Each morning the research agent proposes topics. The marketing lead picks one, the scripting agent drafts it, and the draft waits in the approval workspace. Once a person approves it, the pipeline renders the video and reports when it is ready. The orchestration, built with n8n, keeps each step in order and retries anything that fails.
Manual work removed
- Daily manual topic research
- Writing first-draft scripts from scratch
- Moving files and status updates between tools by hand
- Manually starting each video render
Facing something similar? Let’s map it together.
Related
Next case · Fintech
Automated Credit Onboarding Platform
