LLM-wiki: A Personal Wiki Written and Maintained by LLM
Traditional notes: you write and maintain everything. Typical RAG tools: retrieval starts fresh each time and often does not accumulate. LLM-wiki: LLM writes and maintains the knowledge layer while you focus on thinking and direction.
How Is This Different from RAG or Q&A over Documents?β
| Dimension | Typical RAG / NotebookLM | LLM-wiki |
|---|---|---|
| Storage | Retrieval index over source docs | Source docs plus AI-maintained knowledge pages |
| Query path | Retrieve and compose ad hoc each time | Query maintained pages first, then source docs when needed |
| Accumulation | Limited long-term accumulation | Continuous accumulation of summaries, topics, links |
| Contradiction tracking | Minimal | Periodic checks for contradictions, outdated notes, orphan notes |
| Output persistence | Answer is often one-off | Strong answers can be saved as new notes |
| Data location | Usually cloud | Fully local in your Obsidian vault |
Three Layersβ
| Layer | Content | Owner |
|---|---|---|
| Your source layer | notes, articles, highlights, transcripts | you |
| AI organization layer | summaries, topic pages, comparisons, indexes, logs | AI |
| Rule layer | conventions and operating guidelines | you + AI |
Core idea: you lead strategy and thinking, AI handles routine maintenance.
Three Core Operationsβ
Ingestβ
When you add a new source, AI can:
- read and discuss key points
- write a summary page
- update related topic links
- log the update
Queryβ
AI first consults maintained wiki pages, then pulls source evidence when needed. High-value answers can be saved directly as new notes.
Lintβ
AI periodically checks your wiki for:
- contradictions
- outdated statements
- isolated notes
- missing high-priority topics
Two Special Filesβ
- index.md: navigation and one-line summaries
- log.md: chronological operation log
How GPT AI Flow Supports This Paradigmβ
| LLM-wiki Need | GPT AI Flow Capability |
|---|---|
| AI can read your notes | local semantic indexing over vault content |
| AI can write back | direct read/write to Obsidian files |
| reusable workflows | skill system with one-command workflows |
| proactive rhythm | scheduled prompts and writing rhythm support |
| traceability | version-aware workflow and history support |
| result persistence | save conversation outcomes to notes |
Typical Use Casesβ
- personal growth journaling and reflection
- long-cycle topic research
- reading companion workflows
- product research and competitive analysis
- course notes and hobby deep dives
Privacy and Local-First Designβ
- wiki pages, source docs, and indexes stay on your local disk
- external calls are limited to chosen model interactions and indexing operations
- all notes remain plain markdown and portable
Further Readingβ
Join Usβ
- Experience for free immediately:
- Contact Us

- Contact Email: hello@gptaiflow.com
- Product Feedback:
- Tencent Questionnaire: Click here
- Google Questionnaire: Click here
- π¬ Have a question? Check out the FAQ for quick solutions: Click here
Thank you for choosing GPT AI Flow, together building the essential tools for the super individuals of the future!