Show HN: Setup unlimited accurate agent memory with Past.dev, BEAM10M 85%

Category: ai-ml

Tags: ai-agents, memory, llm, benchmarking, temporal-reasoning

Score: 6.5/10 (Innovation: 7, Technical: 6, Documentation: 7, Utility: 6)

past.dev is a memory layer for AI agents that tracks the temporal validity of facts, replacing similarity-based vector search with time-aware recall. It provides a simple API for ingesting text with timestamps and querying current facts, achieving state-of-the-art results on the BEAM long-context benchmark. The project is notable for its innovative approach to maintaining an evolving knowledge state, but the repository primarily contains benchmark harnesses rather than the core implementation.

Target audience: AI engineers, backend developers building agent applications

Repository: https://past.dev/blog/introducing-past-dev · Python · MIT · 4 stars

View on Hacker News