Seekdb - OceanBase's first open-source AI-native hybrid search database

What is Seekdb?

Seekdb is OceanBase's open-source AI-native database, addressing the pain points of multimodal data fusion and real-time processing in AI application development. It supports fusion queries combining vector retrieval, full-text search, and structured data filtering. A single SQL query can complete cross-modal data processing, such as complex queries like "transactions exceeding 50,000 yuan in the last 7 days, abnormal location, and behavior similar to historical fraud samples," without requiring cross-system calls. It requires only a minimum of 1 CPU core and 2GB of memory, supports one-click installation via pip, and can be embedded in intelligent agents, local applications, or run as a service. Developers can quickly build AI applications with just three lines of code. Based on OceanBase's transaction engine, it ensures real-time index updates after data is written, guaranteeing data consistency and real-time performance in highly sensitive scenarios such as finance and government. Open-source under the Apache 2.0 license, it is compatible with over 30 mainstream AI frameworks such as LangChain and HuggingFace, and provides SQL and Python SDKs to lower the development threshold.

Seekdb's main functions

• Hybrid search capabilitiesIt supports unified retrieval of vector, full-text, scalar, and geospatial data, enabling efficient querying of multimodal data.
• The integration of AI inference and data processingIt enables direct execution of operations such as embedding, reordering, and large language model inference within the database, thereby improving data processing efficiency.
• Lightweight deployment and quick startIt only requires a minimum of 1 CPU core + 2 GB of memory to run, supports one-click installation with pip install, and starts up in just a few seconds.
• Open source and broad compatibilityIt is open source under the Apache 2.0 license, seamlessly compatible with the MySQL ecosystem, and supports more than 30 mainstream AI frameworks such as Hugging Face and LangChain.
• Low latency and high performanceIt can retrieve billions of multimodal data in milliseconds, meeting the needs of scenarios with high real-time requirements.
• Multi-running mode supportIt offers two operating modes: embedded and client/server, flexibly adapting to different development needs.
• Simplify the development processDevelopers only need three lines of code to build AI applications such as knowledge bases and intelligent agents, lowering the development threshold.

SeekDB's technical principles

• AI native architectureSeekdb employs an AI-native design, deeply integrating AI inference with data processing. It supports operations such as embedding, reordering, and large language model inference performed directly within the database, improving overall performance through an optimized execution engine.
• Hybrid search engineIt integrates unified retrieval capabilities for vector, full-text, scalar, and geospatial data. Through a multi-stage retrieval mechanism, it achieves high-precision search with low latency, meeting the complex query needs of multimodal data.
• Lightweight deployment and quick startSeekd is designed to be lightweight, requiring only 1 CPU core and 2 GB of memory to run, and supports one-click installation via pip install, with startup taking only a few seconds.
• Open source and compatibilityIt is open source under the Apache 2.0 license, seamlessly compatible with the MySQL ecosystem, and supports more than 30 mainstream AI frameworks such as Hugging Face and LangChain.
• Low latency and high performanceWith its optimized index structure and retrieval algorithm, SeekDB can respond to multimodal data retrievals of billions of records in milliseconds, making it suitable for scenarios with high real-time requirements.
• Flexible operating modesIt supports both embedded and client/server operating modes to meet different development needs and provide greater flexibility.
• Deep integration of AI frameworkSeekdb deeply integrates with a variety of mainstream AI frameworks, supporting direct vector embedding and model inference within the database, reducing data transmission and processing latency.

Seekdb project address

• Project official websitehttps://www.oceanbase.ai/
• Github repositoryhttps://github.com/oceanbase/seekdb

Application scenarios of Seekdb

• Semantic searchBy leveraging hybrid search capabilities, it can quickly retrieve text content, achieve accurate semantic matching, and improve search efficiency and user experience.
• Knowledge base Q&AIt supports the construction of intelligent knowledge bases, quickly responds to user queries, and provides accurate answers and information, making it suitable for internal knowledge management and service support within enterprises.
• Recommendation systemLeveraging its high performance and multimodal data processing capabilities, it provides personalized recommendations to users, improving the accuracy and real-time nature of the recommendations.
• In-database model inferenceIt supports direct execution of AI model inference within the database, reducing data transmission and processing latency and improving overall performance.
• Multimodal data retrievalIt is suitable for scenarios that require processing multiple data types (such as text, vectors, geospatial data) to achieve unified retrieval and analysis.
• Intelligent customer service and virtual assistantThrough rapid retrieval and semantic understanding, it supports quick responses and accurate answers from intelligent customer service and virtual assistants, improving service efficiency.