A 21-bot content team routes every post through human approval
At a glance
A small content business split research, editing, production, and scheduling across role-specific Grok Bots. This workflow is listed under content operations, multi-agent, human approval.
- Agents / tools
- Grok Bot, Google Drive, Metricool
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) Assign each bot a narrow role, including a general manager, implementation bot, auditor, page researchers, editors, rights checker, designer, video editor, media sourcer, scheduler, and business admins. 2) Keep per-bot instruction files on the shared computer, with team rules and a master post workflow in a shared folder; keep copies and dated instruction snapshots in Google Drive. 3) Pass work between bots and run each post through research, first draft, edit, checks, and the owner's approval before the publishing bot schedules it in Metricool. 4) Put rule changes in the master document, have the bots reread and update their own instructions, then have two checking bots verify the rollout. 5) Require the owner's explicit approval before any post, deletion, or purchase.
How it went
The author said the content quality was good and that the approval gate worked. They also reported about $180 in usage in a day and $325 in extra charges that month.
Worth knowing
The author could not see usage by bot and suspected long chat histories, bot handoffs, and group chats increased costs. The multi-bot setup was still being evaluated.
