Reproduce “PolyMemDB: A Polyglot Database System for AI Memory Management” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Yu Wang, Jiaheng Lu

PolyMemDB answers questions about facts in long conversations by routing different data to suitable stores and retaining dated observations with confidence and polarity. It uses those records to assess conflicting evidence and expose support for an answer. In a LongMemEval demonstration spanning 48 sessions, it answered a book-discount question with 20% and showed supporting records. This case illustrates evidence traceability; the paper reports no controlled benchmark showing improved accuracy or general reduction in hallucinations.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.