StackMap
Subscribe

PageIndex vs supavec

Vectorless, reasoning-based RAG — builds a hierarchical tree index from long documents so an LLM retrieves by relevance instead of similarity. No chunking, no embeddings, no vector DB. — versus — Open-source RAG-as-a-service (the Carbon.ai alternative): upload any data source, get vector search and a chat API in minutes — Supabase-based, multi-tenant with RLS, streaming responses.

The curated verdict

Both answer questions over your documents; supavec is vector-DB RAG-as-a-service (embed + similarity search), PageIndex is vectorless — a reasoning tree index with no chunking or embeddings.

PageIndexsupavec
Stars35k1.2k
Forks3.0k104
LanguagePythonTypeScript
LicenseMITApache-2.0
Last activity2 days ago7 months ago
Topicsragrag
Curated connections53

PageIndex — the curator's take

Reach for it on long, structured professional documents (contracts, filings, manuals) where similarity search returns 'similar but irrelevant' passages and you need reasoning over document structure. Tradeoff is per-query LLM reasoning cost/latency versus a cheap vector lookup, and you still need clean parsed text upstream. If your corpus is huge, homogeneous and similarity is good enough, a vector DB (chroma, supavec) is cheaper.

supavec — the curator's take

The API-first take on RAG: POST a file, query a /chat endpoint, done — with genuinely production-minded internals (row-level security for tenant isolation, batched embeddings cutting OpenAI cost ~65%, Redis sliding-window rate limits, request tracing). Use it when you want RAG behind an API for your product without assembling the pipeline yourself. Know the shape: it's an open-core SaaS — usage-tiered billing on the cloud version, and self-hosting means running the Next.js + Supabase + Upstash stack yourself with docs that assume you'll read the code. For a self-contained enterprise RAG product with a UI, MaxKB; for just the vector store, Chroma.