The problem nobody talks about $100B+ is wasted annually in research duplication. Not because researchers are lazy or careless. doesn't work — without asking for extra effort. The root cause is publication bias. Labs fund projects based on partial or incorrect assumptions The consequences are concrete: If lab A doesn't publish the failure of formula X, This happens every day. Across every scientific discipline. The failed hypothesis at 11pm. This knowledge disappears. Into lab notebooks nobody reads, Every existing tool — Notion, Obsidian, Roam — asks for after the work is done. What if the research process left a trace automatically, That's MemoryGraph. Every unit of thought lives as a typed node in a personal Observation — something you noticed Hypothesis — something you believe might be true Conclusion — something you've established DeadEnd — something that didn't work OpenQuestion — something you don't know yet Each node carries a full temporal history — every change The graph is never a snapshot. It is a recording. Git MemoryGraph Repository Personal knowledge graph Commit NodeState — a belief at a moment in time Fork SubgraphToken — a signed copy of selected nodes Diff Semantic delta between two trajectories Pull request MergeProposal with conflict detection When two researchers need to share knowledge, one issues SubgraphToken — a signed, scoped selection of nodes. The dark matter of research finally has a place to live. ✅ Graph store with Kuzu (embedded, zero infra) ✅ Full node versioning — nothing ever deleted ✅ Memory Agent — LLM-powered entity extraction ✅ Quality gate before every graph write ✅ Contradiction detection ✅ Link Agent — semantic edge suggestion ✅ LLM-agnostic — works with Anthropic, OpenAI, or local Language: Python 3.11+ Graph DB: Kuzu (embedded) LLM: agnostic — any model via structured prompting License: AGPL-3.0 Researchers who want to try it on a real project Builders who want to implement any phase of the roadmap Critics who want to find the failure modes → github.com/alteavane/memory-graph Open an issue. Fork the repo. Break the design. The goal is not consensus — it's the best possible system.