Antfly packs BM25, vectors, graph traversal
Antfly is a distributed search engine built in Go on top of etcd’s raft library, designed to handle the whole AI retrieval stack in one system. It combines full-text BM25 search, vector similarity, graph traversal, and multimodal indexing for text, images, audio, and video, while also generating embeddings, chunking, and graph edges automatically. The project also includes built-in RAG agents, local ML inference, reranking, and a dashboard so teams can move from raw data to searchable knowledge without stitching together a pile of separate services.
Hot take: this feels less like “just another vector DB” and more like an attempt to collapse an entire RAG memory pipeline into one distributed engine.
- –The strongest angle is the breadth: lexical, semantic, graph, and multimodal retrieval are all first-class instead of bolted on.
- –The Go + raft foundation makes it feel practical for self-hosted infra teams that want one binary-ish stack with real distributed behavior.
- –Built-in ML inference and RAG tooling are a meaningful differentiator if the project keeps the developer experience simple.
- –The risk is scope: systems that try to own search, graph, embeddings, and orchestration can become impressive demos but hard-to-adopt platforms.
DISCOVERED
21d ago
2026-03-21
PUBLISHED
21d ago
2026-03-21
RELEVANCE
AUTHOR
Github Awesome