The open-source, AI-native vector database that became the retrieval and memory layer for production GenAI apps
HQ: Amsterdam, Netherlands (remote-first; US entity Weaviate LLC, press releases datelined Amsterdam and New York) | Founded: 2019 | Employees: ~90–100 (company states 90+; Crunchbase 51–100; PitchBook 99) | Stage: Series B (Private) | Website: https://weaviate.io
Weaviate is the company behind the open-source Weaviate vector database, written in Go and built to serve search, RAG and agent workloads from a single system. The project started in Amsterdam as SeMI Technologies, spun out of ING Labs, with Bob van Luijt as CEO, Etienne Dilocker as CTO and Micha Verhagen as COO; van Luijt dates the open-source launch to March 2019. The legal entity is Weaviate B.V. in Amsterdam, with Weaviate LLC as the US entity. The company is remote-first, and 2026 press releases are datelined from both Amsterdam and New York.
Unlike a bare similarity index, Weaviate stores objects, vectors and a keyword index together, so hybrid search (dense plus BM25), metadata filtering, sorting and multi-tenancy are first-class — the properties that matter once a prototype becomes a multi-tenant product. The commercial layer is Weaviate Cloud, running on AWS, Google Cloud and Azure (about 40 regions on dedicated plans), with SOC 2 Type II attestation and HIPAA-eligible Enterprise Cloud on AWS. Around it sit Query Agent for natural-language retrieval, Weaviate Embeddings for hosted vectorization, and Engram, a managed memory service for agents that reached general availability in June 2026.
Funding totals roughly $67.7M: a $1.6M seed in August 2020 (Zetta Venture Partners with ING Ventures), a $16M Series A in February 2022 co-led by NEA and Cortical Ventures, and a $50M Series B in April 2023 led by Index Ventures with Battery Ventures, plus an undisclosed strategic investment from Ricoh in March 2026. The company states 90+ employees, and its Global VP of Sales has claimed 1,800+ paid customers, 60 of them enterprises in production. Named case studies include Booking.com, Morningstar, Instabase, DocsBot and Stack AI.
Weaviate's bet was that search for AI applications needs more than a nearest-neighbour index: objects, dense vectors, keyword (BM25) scoring and filters must live in one system, or every team ends up gluing a vector store to a database to a cache. That design — hybrid search by default, native multi-tenancy, pluggable vectorizers for OpenAI, Cohere, HuggingFace and Voyage, and deployment from a laptop to BYOC on AWS/GCP/Azure — is why Booking.com replaced a tuned OpenSearch deployment with Weaviate and reported 20x faster performance at roughly 40% lower cost, and why DocsBot runs 50,000+ isolated tenants on a single cluster.
The 2025–2026 roadmap is a deliberate move up the stack as raw vector search commoditizes. Query Agent (general availability September 2025) turns natural language into multi-collection retrieval with decomposition and reranking; Engram (GA June 2026) sells agent memory as managed infrastructure; Agent Skills wires Weaviate into Claude Code, Cursor, GitHub Copilot and Gemini CLI. GigaOm's Leader-and-Outperformer placement (announced January 2026) and a Gartner Emerging Leader rating are third-party validation of that shift from index to platform.
The uncomfortable numbers sit underneath. There has been no announced priced round since the $50M Series B of April 2023 — more than three and a half years — while Qdrant raised $50M in March 2026, Pinecone has $138M+ in the bank and Supabase now trades at a $10B valuation. Revenue is not disclosed, and third-party 2024 estimates range from $12.3M (Latka) to $37.5M (UpMarket citing Sacra). With roughly 90 employees, 1,800+ paid customers (company claim) and a strategic investor in Ricoh, Weaviate is credible and shipping fast, but undercapitalized relative to peers — and its own open-source core is the cheapest alternative to paying it.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki