Bilingual

Offline knowledge, buses, and note-taking
a day ago
- Knowledge stored only in memory is risky because humans are forgetful, while computers excel at remembering.
- Documentation reduces the 'bus factor' by ensuring multiple people can perform tasks, spreading critical knowledge.
- A personal documentation system should capture tasks quickly, show immediate priorities, and be always accessible.
- Effective note-taking frees mental space for creative problem-solving rather than remembering tasks.
Why organizing knowledge in the age of AI sucks, and how I solve it
14 days ago
- The author spends significant time in AI chat sessions for personal and professional tasks, but faces frustrations in retrieving and reusing context from past conversations.
- Current tools like Claude Projects, linear issues, or doc platforms fail to align with human mental models for organizing knowledge, leading to inefficiencies in storing and loading AI context.
- Key insights include the mismatch between session-based organization and human memory, the lack of meta-context in AI readings, and the overlap between human and AI memory needs for natural language artifacts.
- The author built a minimal solution using Claude Code skills to create a knowledge base with folders as topics and files as items, enabling better context capture and retrieval.
- Future plans involve developing an Obsidian-like view layer and more granular control over context loading to enhance the system.
You Have an AI Working Agreement. Write It Down
17 days ago
- An AI Working Agreement consolidates a team's scattered AI usage rules into a single document, detailing delegation, human oversight, review processes, data boundaries, ownership, and amendment procedures.
- Teams already possess implicit AI rules through habits and templates; codifying them prevents knowledge loss during employee turnover and aids new member onboarding.
- Key components include standing defaults, hard boundaries (data, judgment, relationship), active templates, disclosure norms, concern escalation paths, and review triggers.
- Creating the agreement involves gathering existing decisions, filling gaps, assigning ownership, setting amendment rules, and choosing a central location for accessibility.
- Benefits include reducing single points of failure, enabling consistent decision-making, ensuring compliance with organizational policies, and facilitating team alignment and trust in AI outputs.
Karpathy, Google, Tan agree Markdown is the answer, but not for the same problem
22 days ago
- Andrej Karpathy's 'LLM Wiki' proposes using linked Markdown files for AI agent memory.
- Google introduced Open Knowledge Format (OKF v0.1), packaging organizational knowledge in Markdown for enterprise use.
- Garry Tan's gstack uses Markdown files to define specialist coding agent roles, achieving high GitHub popularity.
- Markdown has become the standard format for agents to read and write, evolving from developer conventions like CLAUDE.md.
- Competitive advantage shifts from model ownership to owning accumulated Markdown knowledge bases that are portable across systems.
- The durability of Markdown standards like OKF depends on adoption and ecosystem development to avoid remaining just ideas.

Hasty Briefsbeta

knowledge management

Offline knowledge, buses, and note-taking

Why organizing knowledge in the age of AI sucks, and how I solve it

You Have an AI Working Agreement. Write It Down

Karpathy, Google, Tan agree Markdown is the answer, but not for the same problem