As AI advances and enters production, the real-time data warehouse Hologres continues to evolve its AI capabilities. It supports an end-to-end workflow from data mining and parsing to intelligent inference and search, handling unstructured, structured, and semi-structured data. By applying AI, Hologres unlocks greater value from your data and drives fine-grained business growth.
AI Models and AI Functions
Hologres provides a set of mainstream large language models (LLMs) from Model Studio or built-in, GPU-powered Hologres AI nodes. You can deploy models from the console with a single click and then use AI Functions in Hologres to call them with SQL. Examples include ai_embed and ai_gen. AI Functions automatically route requests to the deployed models. The entire workflow runs in SQL, eliminating the need for extra code, separate AI service deployment, or data exports. This provides a complete, out-of-the-box AI workflow for storage, retrieval, and inference in a single location where data and models are managed together. For more information, see AI Models and AI Functions.
Unstructured Data Object Tables
In AI scenarios, unstructured data—such as text, images, audio, and video—captures behavior patterns and complex semantics in diverse forms. Hologres 4.0 and later supports Object Tables, which allow you to access unstructured data and its metadata stored in an OSS data lake as tables. When combined with AI Functions, Object Tables support automatic embedding and segmentation of unstructured data. The results are stored in a structured format within your Hologres instance, which enables vector search and full-text search. This allows your enterprise to explore, retrieve, and analyze unstructured data in addition to structured and semi-structured data. This capability expands your data discovery scope and delivers more fine-grained business value. For more information, see Unstructured Data (Object Tables).
Vector Search
Hologres 4.0 upgrades its vector search capability. It uses the HGraph vector search algorithm and supports hybrid memory and disk indexing. It can handle trillion-scale vector data for both ingestion and recall, delivering high performance for recall speed, recall accuracy, and index build time. Hologres vector computation integrates AI resources and AI Functions to support end-to-end multimodal data transformation, vectorization, ingestion, and analysis. You can use it for applications such as similarity search, image retrieval, and scene recognition. For more information, see Vector Computation Overview.
Full-Text Search
Hologres 4.0 uses the high-performance full-text search engine Tantivy and the BM25 algorithm for fast, accurate full-text search. It supports a rich set of tokenizers and lets you build full-text inverted indexes for multiple languages. It also supports several search modes, such as keyword match, phrase search, natural language search, and term search, allowing you to choose the mode that best fits your business goals. Combining full-text and vector search can boost AI performance in text-based retrieval tasks, such as retrieval-augmented generation (RAG).
MCP
Hologres implements the Hologres MCP Server based on the MCP protocol. This server acts as a universal interface between AI agents and the Hologres database, enabling seamless communication. This allows AI agents to retrieve Hologres metadata and run SQL operations. Hologres supports multi-channel deployment of the MCP Server and ChatBI Agent. For more information, see MCP and Chat BI.
Rich AI Ecosystem Integrations
Hologres offers a rich set of AI ecosystem integrations:
-
Integration with Dify: Dify is an open source large language model (LLM) application development platform. It combines Backend-as-a-Service (BaaS) and LLMOps principles to help developers quickly build production-grade generative AI applications. The hologres_text2data plug-in is now available on the Dify official marketplace, and its source code is available on GitHub. You can use Hologres with Dify to quickly build enterprise ChatBI applications. For more information, see Build a ChatBI Application Fast Using Dify and Hologres.
-
Python SDK: Hologres is compatible with PostgreSQL 11. You can use Psycopg to access Hologres and achieve high-performance data reads and writes. Hologres also provides a vector-enabled Python SDK for high-performance, low-latency vector computation. For more information, see Search SDK.