
CHOI builds LLM Wiki workflow system
AI creator CHOI built an operational workflow system centered around Andrej Karpathy's 'LLM Wiki' pattern to capture employee responsibilities and company context in structured markdown. The system automatically decomposes commands into specialized agent tasks, prompting the creator to declare humans as the primary organizational bottleneck.
While automating knowledge base updates and mapping workflows makes perfect sense, declaring humans as the "bottleneck" ignores the fact that AI still depends on human operational reality, and actual execution requires human decision-making that can't be fully automated in static wikis.
- –The LLM Wiki pattern shifts the burden of documentation upkeep from humans to AI, solving the classic problem of outdated company wikis.
- –Centering an organization's operations around an LLM-accessible markdown wiki allows autonomous agents to coordinate complex inter-departmental tasks with high context.
- –Declaring humans as the "bottleneck" highlights the massive discrepancy between the speed of AI context processing and the rate at which human teams digest and act on information.
- –A major challenge will be handling the delta between the documented wiki workflows and the real-world ad-hoc actions employees take daily.
DISCOVERED
1d ago
2026-07-07
PUBLISHED
1d ago
2026-07-07
RELEVANCE
AUTHOR
arrakis_ai