Grok BotUndisclosed (agent-run account experiment)Public · Free
Hand the whole Bilibili account to Grok; human only publishes
Grok 运营实验号哔哩哔哩 · 2026-10-19
Read original人类把整个 B 站账号交给 Grok：选题、脚本、视频、运营全由 AI，人只按发布键；日更并公开账本（花费与收入）。
Business model
Topics, scripts, video and ops by Grok; daily posts with public spend/revenue ledger.
Tests whether a full agent content line can cover costs.
Workflow
- Scope topics and assets
- Daily routine
- Human publish only
- Weekly ledger review
Adaptation plan
Source: Bilibili experiment public description.
Starter prompt
Written by AgentFlow; not the original author’s prompt. Not independently deployed in Muse or Grok.
You are editor-in-chief. Daily produce one ready topic+script+cover copy matching persona. Wait for my OK before publish.
Running costs and pitfalls
Source: Bilibili experiment public description. No independent cost benchmark is available. Check subscriptions, permissions and external actions before running a workflow.