Remembrane is a new, local-first persistent memory solution for AI agents designed to overcome the limitations of cloud-based or framework-heavy alternatives. It stores an agent's entire memory in a single SQLite file with no required dependencies, offering deterministic behavior and testability. Remembrane features a unique hybrid recall system that combines vector and BM25 keyword search for exact, explainable ranking, incorporating factors like recency, importance, and learned usefulness from task outcomes. The system also supports time-travel through memory history and conflict-aware recall, with adapters available for popular frameworks like LangChain and CrewAI. AI
IMPACT Provides a novel, self-contained memory solution for AI agents, potentially simplifying development and deployment.
RANK_REASON The item describes a new software tool for AI agents, not a frontier release, significant industry move, or academic research.
Read on HN — claude cli stories →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
