AI Briefing#bm25
The latest AI and developer news about #bm25, with the original source and a short summary.
The latest AI and developer news about #bm25, with the original source and a short summary.
Achieved 99% reduction in context tokens compared to grep on 1 million records with 98.5% hit@1 accuracy.
The MIT-licensed tool uses SQLite FTS5 and optional local embeddings to provide offline, Markdown-based memory for AI coding agents.
ParadeDB optimized Tantivy to close an initial 8x performance gap with PlanetScale's TIN, achieving higher throughput in fair BM25 benchmarks by fixing data locality and algorithm selection.
Knowledge Agents injected with specialized knowledge outperform general-purpose frontier models in specific domains.
The open-source tool Ratel uses the BM25 algorithm to dramatically reduce MCP tool context usage in Claude Code.
LLM Wiki v2 grows into an operational knowledge base by adding lifecycle management, graphs, and automation.
OpenSquilla has unveiled an open-source AI agent runtime that reduces token costs.
It raises the need for a dedicated agent model that orchestrates the entire search process.
mind-mem has released a hybrid memory MCP combining BM25+vectors+RRF.
Attaching an agent to simple BM25/embedding search raised NDCG up to 0.453.
AI Search provides vector-keyword hybrid search and per-instance search.
By attaching CPT and BM25 RAG to Qwen3-30B-A3B, the team won the Excellence Award while running in a 24GB environment.
Dropbox Dash reduces tokens and latency through index, knowledge graph, and MCP optimization.
Contextual Retrieval leverages contextual information to improve RAG retrieval failure rates by up to 67%.
Attaching document context before and after chunks to build embeddings and BM25 can significantly reduce RAG retrieval failures.
It presents a strategy for building an agent that handles 1M context using an 8k context model, and using this to train long-context models.