Neon rebuilds its docs search entirely on its own stack: Lakebase Search, Functions, AI Gateway
matei_zaharia · x · 2026-10-27
Neon replaced its hosted search vendor and rebuilt neon.com docs search fully on its own primitives: Lakebase Search extensions, Neon Functions, Neon AI Gateway, and Lakebase Postgres.
Key components:
- lakebasevector: pgvector-compatible, with an index needing a fraction of HNSW memory (under 10 GB for 100M vectors vs 300 GB for HNSW), scaling past a billion vectors
- lakebasetext: brings BM25 to Postgres, weighing term rarity and document length unlike native tsrank, returning top results without scoring every match
Both ship new index types (lakebaseann, lakebasebm25) designed for Neon's compute/storage-separated architecture. A full dogfooding writeup of hybrid vector+BM25 search in Postgres.
More from Infra
- Anthropic S-1 reveals $800B+ in off-books future compute commitments — GaryMarcus · 2026-10-27
- ~42% of Anthropic's ~11GW contracted compute is NVIDIA GPUs — ns123abc · 2026-10-27
- EdgeRock-4B holds KV cache at 72 MiB across 256K context, 99.1% smaller than Qwen3.5-4B — alifcoder · 2026-10-27
- RTX 5080 price jumps from $1658 to $2318 in ten days on Amazon — Frizzy-MacDrizzle · 2026-10-27
- ComfyUI on 12GB VRAM: measured peaks show offloading turns 24s into 11 minutes — Impossible-Mail-4619 · 2026-10-27
- PyTorch details 2.12x faster on-device streaming ASR on Arm SME2 via ExecuTorch — PyTorch · 2026-10-27