Concept — Knowledge Compounding
The core value proposition of the LLM Wiki: the wiki is a compounding artifact, not a cache that resets on every query.
What It Means
In the RAG pattern, every query re-fetches and reassembles from raw — knowledge does not accumulate. In the LLM Wiki, cross-references are already there, contradictions are already flagged, and the synthesis already reflects everything you’ve read. Every source added and every question asked makes the wiki richer and more coherent.
Why It’s Key
The most tedious part of maintaining a knowledge base is bookkeeping: updating cross-references, keeping summaries current, flagging where new data overturns old claims, maintaining consistency across dozens of pages. People abandon wikis because the maintenance burden grows faster than the value.
The LLM never tires, never forgets to update cross-references, and can touch 15 files in one pass. With near-zero maintenance cost, the wiki actually sticks around — and that is the condition under which compounding can happen.
Two Sources of Compounding
- Ingest compounding: each ingest integrates the new source into the existing network rather than archiving it in isolation.
- Inquiry compounding: good query answers are saved back to the wiki as new pages, so the inquiry itself accumulates.
Related
- llm-wiki-pattern — the difference from RAG
- three-operations — how ingest/query create compounding
- three-layer-architecture — the persistent wiki layer is the vessel of compounding