Full-text search with BM25
Create inverted indexes over text columns and query them with the `@@` operator, phrase constructors, and BM25 ranking.
SereneDB
Freemium
SereneDB is a search-OLAP database for real-time full-text, vector, hybrid search, and analytical queries over the same data. It uses SQL and the PostgreSQL wire protocol, helping developers connect with familiar clients while keeping search and analytics in one engine.
What is SereneDB?
SereneDB is an open-source search-OLAP database for real-time search analytics. It combines full-text, vector, hybrid, and analytical querying in one SQL-accessible engine, allowing search predicates, semantic ranking, filtering, and aggregations to operate over the same data.
The product uses the PostgreSQL wire protocol and a PostgreSQL-compatible SQL dialect, so it can be accessed with psql and other compatible clients and drivers. It ships as a single binary, with Docker and Linux installation paths plus downloadable builds for macOS and Windows.
SereneDB also supports reading Parquet, CSV, and JSON from object storage, HTTP, and the Hugging Face Hub. External files can be indexed in place for zero-ETL search, while SQL functions can generate embeddings for vector and hybrid retrieval workflows.
What can SereneDB do?
Vector and hybrid search
Store embedding vectors beside relational data, build IVF indexes, and combine lexical filtering with semantic ranking in a single query.
Search and analytics on shared data
Run SQL aggregations, filters, and analytical queries over search results without copying data into a separate warehouse or search system.
PostgreSQL-compatible access
Connect through the PostgreSQL wire protocol using `psql` or PostgreSQL-compatible clients and drivers. The SQL dialect closely follows PostgreSQL conventions.
Remote and external-data querying
Read Parquet, CSV, and JSON from object storage, HTTP, and the Hugging Face Hub, or index supported external files in place for zero-ETL search.
SQL embedding generation
Use the `ai_embed` function to generate vectors through a configured provider, including OpenAI-compatible endpoints and local Ollama servers.
Use Cases
“Documentation and knowledge-base search”
Index documentation or other text collections, then use full-text, vector, or hybrid retrieval to find relevant passages and rank results by keyword match or meaning.
“Retrieval for AI applications”
Generate embeddings in SQL and use vector or hybrid search as a retrieval layer for grounded answers and agent-oriented database workflows.
“Real-time search analytics”
Filter search results and aggregate them in the same statement, such as grouping matching records and calculating averages over the selected set.
“Data-lake and remote-file search”
Query datasets stored as Parquet, CSV, or JSON in object storage, HTTP locations, or the Hugging Face Hub, including in-place indexing where supported.
“PostgreSQL-based application search”
Add search capabilities to applications that already use PostgreSQL-compatible drivers, while retaining SQL access to relational data and analytical operations.
Frequently Asked Questions
How do I install SereneDB?
The documented quick start supports Docker and Linux installation scripts. The download page also lists Linux packages and archives, plus macOS DMG and Windows EXE downloads. Other installation options are available through the GitHub releases page.
How do I connect to a running instance?
SereneDB speaks the PostgreSQL wire protocol. The quick start shows connecting with `psql -h localhost -p 7890`; no credentials are required by default in that local setup.
Can SereneDB query data without a traditional import job?
Yes. The documentation shows reading Parquet, CSV, and JSON from object storage, HTTP, and the Hugging Face Hub through SQL. It also documents indexing external files in place for zero-ETL search.
Can one query combine search and analytics?
Yes. A documented workflow uses a full-text predicate to select matching rows and then applies `GROUP BY`, ordering, and aggregate functions in the same SQL statement.
Which embedding providers can be used with `ai_embed`?
The quick-start documentation shows an OpenAI-compatible provider and states that a local Ollama server can also be used. The provider is configured with `CREATE SECRET`.
Quick Facts
- Product type
- Open-source search-OLAP database
- Primary interface
- SQL over the PostgreSQL wire protocol
- Search modes
- Full-text, vector, and hybrid search
- Deployment options
- Docker, Linux, macOS, and Windows downloads
- Data sources
- Tables plus Parquet, CSV, and JSON from remote locations
- Default quick-start port
- 7890
SereneDB Alternatives
Vespa.ai
vespa.ai
Vespa is an AI search platform for building search, retrieval-augmented generation, recommendation, personalization, and agent applications over text, vectors, tensors, and structured data. It is designed for developer teams that need configurable ranking and distributed operation at production scale.
Supabase
supabase.com
Supabase is a Postgres development platform for building applications with a database, authentication, APIs, realtime features, serverless functions, storage, and vector support. It helps developers move from application code to a managed backend through a dashboard, client libraries, CLI, and generated APIs.
Upstash Vector
upstash.com
Upstash Vector is a serverless vector database for storing and querying embeddings and metadata in AI and machine-learning applications. It supports dense, sparse, and hybrid indexes through a REST API and SDKs for Python, TypeScript, Go, and PHP.
Powabase
powabase.ai
Backend platform for AI apps with Postgres, auth, storage, realtime, retrieval, and agents
Weaviate
weaviate.io
Weaviate is a vector database for building AI-native applications, including retrieval-augmented generation systems. It helps developers store and retrieve vector data, connect applications to language models, and deploy the database in a managed, self-hosted, or private-cloud environment.
Spice AI
spice.ai
Open-source SQL query and hybrid search engine for data-intensive apps and AI agents, with zero ETL.