What it is
Weaviate is an open-source vector database for searching text, images and other data by meaning. It blends vector and keyword search in a single query and can create embeddings for you by connecting to providers like OpenAI or Cohere. Self-host it with Docker or Kubernetes, or use Weaviate Cloud, the managed version.
How to use it
- Self-host it with Docker or Kubernetes, or spin it up on Weaviate Cloud.
- Let it create embeddings on import, with modules for OpenAI, Cohere, Hugging Face and more.
- Run hybrid queries that mix meaning and keywords.
- Isolate each customer with multi-tenancy inside one collection.
- Generate answers (RAG) right from the query with a language model.
What it's good for
- An HR SaaS searching résumés by skills, not just exact keywords.
- A team running Dify that needs a self-hosted vector store for its agents.
- A company with sensitive data that wants semantic search without sending documents to an outside cloud.
Pros
- Open source and self-hostable, or managed on Weaviate Cloud
- Hybrid search (vectors plus keywords) out of the box
- Creates embeddings on import, with no extra pipeline code
- Multi-tenancy built for SaaS products with many customers
Cons
- Running it at scale means tuning memory and replicas
- The free cloud sandbox only lasts 14 days
- Console and docs are English-only
