AI-Native Database Service is a one-stop intelligent data development platform that delivers end-to-end AI-native data services spanning data preparation, data access, knowledge management, and agent development. The service helps enterprises rapidly build and deploy data-driven AI applications.
Overview
AI-Native Database Service is designed for enterprise AI implementation scenarios. The service deeply integrates data management capabilities with AI technology and provides a comprehensive platform that covers data preparation, data access, asset management, agent development, and intelligent applications. Through a progressive four-layer architecture, AI-Native Database Service equips enterprises with full-pipeline capabilities from foundational data management to intelligent data applications.
AI-Native Database Service includes the following core capabilities:
Agent Data Gateway: An enterprise-grade data access security infrastructure for the Agent era. Deployed between all AI agents and enterprise data sources, Agent Data Gateway provides unified secure access, DataWiki-based business semantic understanding, full-link security protection, and production stability assurance. This ensures that any agent can use enterprise data in a secure, controlled, and auditable manner.
AI asset and enterprise knowledge services: Built around OneMeta, this capability provides multimodal data asset and knowledge management, including unified metadata management, industry knowledge accumulation, and multimodal knowledge services for multi-agent architectures.
AI Data Preparation: An intelligent data preparation service built for agent-ready data. AI-driven multimodal connectors automatically acquire, parse, and process data from multiple sources.
Enterprise-grade database Agent Developer Platform: Three development paradigms are supported: Custom Development, intelligent development, and AI-native development. The platform supports low-code visual orchestration and agent-native Skill/MCP development.
Data Agents: A set of ready-to-use intelligent data assistants that cover core scenarios such as data analytics, database O&M, forecasting, and data ingestion.
Architecture
AI-Native Database Service uses a progressive architecture that builds from foundational data management up to intelligent applications:
Layer | Description |
Data Agents | Ready-to-use intelligent assistants for end users, including Analytics Agent (data insights), DAS Agent (database O&M), Forecast Agent (deduction and forecasting), DataBridge Agent (data ingestion), and Data Skill Market (value-added service marketplace). |
Enterprise-grade database Agent Developer Platform | Three development paradigms are supported: Custom Development (Dify + RAGFlow + Supabase), intelligent development (Proactive Agent), and AI-native development (Skill/MCP). |
Agent-native data foundation | Core data capabilities for agents, including multimodal data preparation (multimodal data source ingestion, data parsing, and enrichment) and enterprise data assets and knowledge management (AI-powered knowledge acquisition and generation, domain-wide access control, and intelligent knowledge services). |
Agent Data Gateway | A unified secure access layer deployed between agents and enterprise data sources. Agent Data Gateway serves as the single entry point for all Agent data access. It provides the following capabilities: agent identity and API Key management, fine-grained permission control (database, table, column, and row levels), a security rule engine (Allow/Ask/Notify/Approve/Deny), SQL and API risk prevention, automatic sensitive data masking, specification review and deterministic execution, and full-link auditing with emergency circuit-breaking. |
Cross-cloud multimodal data management | Unified management of diverse data sources, including relational databases (RDS, PolarDB), NoSQL databases (Tair, MongoDB, Lindorm), data warehouses (AnalyticDB, ClickHouse, SelectDB), and cross-cloud access for multi-cloud databases. |
Core modules
Agent Data Gateway
Agent Data Gateway is a secure access infrastructure deployed between all AI agents and enterprise data sources. It provides unified access control, business semantic understanding, full-link security protection, and production stability assurance. Core capabilities include:
Unified secure access: Supports nearly one hundred types of data sources and mainstream Agent clients (MCP/Skill/CLI/IDE plug-ins). Each Agent is assigned an independent identity with row-level and column-level permissions, while database credentials are physically isolated and never leave the gateway.
DataWiki business semantic understanding: Captures field definitions, metric specifications, business rules, and usage experience. Supports intent-driven interaction and clarification protocols to ensure agents accurately understand business context.
Full-link security protection: Automatically masks sensitive data, blocks unauthorized access by default, triggers real-time alerts for anomalous behavior, and provides full-link auditing traceable to the specific Agent, responsible party, intent, and SQL statement.
Production stability assurance: Uses Spec pre-checks, deterministic execution, and a lock-free change engine, building on over a decade of SQL security rules and four-dimensional risk identification capabilities from Alibaba Cloud AI-Native Database Service.
AI asset and enterprise knowledge services
Built around OneMeta, this module establishes a competitive advantage in multimodal data asset knowledge management and extends the boundaries of unified knowledge management and services across vertical industries and the broader internet. Core capabilities include:
Unified metadata management (OneMeta): Provides automated collection and parsing of domain-wide, multimodal data assets, intelligent data asset inventory and business semantic generation, and an enterprise-grade unified data semantic foundation.
Industry knowledge and SOP accumulation: Combines intelligent search and DeepResearch services to continuously collect and summarize expert knowledge across industry, Agent, and context dimensions, reducing the cold-start barrier for agents.
Multimodal knowledge services for Multi-Agent architectures: Integrates vector, knowledge graph, and relational architectures to deliver Multi-Agent-friendly knowledge generation and service capabilities.
Knowledge management agent (Meta Agent): An automated intelligent agent purpose-built for Knowledge as a Service (KaaS). Meta Agent proactively and continuously acquires enterprise knowledge to generate domain-wide data and business knowledge.
AI Data Preparation
An intelligent data preparation service built for Agent-Ready data. After you describe your data preparation objectives, the system automatically acquires and processes data from multiple sources to produce Agent-Ready data that is immediately consumable. Core capabilities include:
AI-driven multimodal connectors: Support continuous real-time data acquisition and write-back from diverse sources including the internet, databases, and documents, with Vibe Coding-based adaptive data connection.
Multimodal Data Preparation: Driven by data preparation task objectives, the system uses adaptive reasoning for processing logic. It automatically determines which data to acquire and how to process it, producing vector data and semantic metadata.
Extensible skillsets and multi-Agent collaboration: Features an open Agent Skill design that supports custom skills for different industries and scenarios, with MCP-based multi-Agent collaboration.
Enterprise-grade database Agent Developer Platform
The platform continuously refines low-code and no-code AI development experiences with Agent technology and builds core capability modules for Agent-native skills, MCP, memory, and knowledge retrieval. Three development paradigms are available:
Custom Development: Deploy enterprise-grade Dify and RAGFlow platforms with a single click, combined with Supabase for high-availability services. This paradigm is suited for enterprise customization scenarios.
Intelligent development: Proactive Agent leverages enterprise data and business semantics to proactively recommend actionable Agent solutions, with support for Vibe Coding to customize agents.
AI-native development: Provides DataSkill and MCP services for secure data access and management, supports the Managed Data Agent hosting model, and enables employee-specific Data Agents for enterprise scenarios.
Data Agents
A family of premium database Data Agents that provides ready-to-use, highly competitive vertical Data Agent products with enterprise-grade solutions for key business scenarios.
Product | Positioning | Core capabilities |
Analytics Agent | A digital employee for data scenarios | Data engineering, data insights, and business recommendations |
DAS Agent | Intelligent database O&M and security | Proactive architecture governance, intelligent fault diagnosis, deep SQL optimization, and security protection and response |
Forecast Agent | Deduction and forecasting | Trend forecasting, demand forecasting, and public opinion prediction |
DataBridge Agent | Internet data and multimodal data acquisition | Data ingestion, file parsing, and data transmission |
Data Skill Market | Value-added service and Agent capability marketplace | Subscribable Agent services and plug-and-play Skill packages |
Scenarios
Secure data access for agents: Multiple teams across an enterprise use clients such as Cursor, Qoder, Claude Code, and Coze to access production data. Through Agent Data Gateway, all agents receive independent identities, row-level and column-level permissions, and automatic sensitive data masking through the unified gateway. High-risk operations undergo Spec review before deterministic execution, and full-link auditing traces each action to the specific Agent and responsible party.
Intelligent data insights: Analytics Agent enables natural language interactive data analysis. Combined with OneMeta, it automatically inventories data assets and generates business semantics to improve data analysis efficiency.
Intelligent O&M: DAS Agent enables unified intelligent O&M across databases, supporting proactive architecture governance, real-time prevention and control, proactive optimization, and emergency response. This helps DBAs transition from firefighting to the role of digital intelligent architects.
Enterprise knowledge management: Multimodal data preparation consolidates documents, meeting recordings, emails, and other multimodal data. Combined with Meta Agent, it builds an enterprise knowledge graph and semantic search engine, enabling knowledge Q&A and cross-Agent knowledge sharing through Managed Data Agent.
Multimodal Data Preparation: DataBridge Agent enables ready-to-use data ingestion with support for skills such as Web Channel, DB Channel, and File Channel to acquire multimodal data in real time and parse documents, producing Agent-Ready data for AI applications.
Benefits
Benefit | Description |
Comprehensive capabilities | End-to-end AI-native data services cover data preparation, data access, knowledge management, and Agent development. A single platform meets all data requirements for enterprise AI implementation. |
Stability and reliability | AI-powered full-link intelligent O&M capabilities support proactive architecture governance, real-time prevention and control, and proactive optimization for efficient and precise database stability assurance. |
Security and trust | Agent Data Gateway serves as the unified entry point for all Agent data access, providing independent Agent identities, row-level and column-level permissions, a security rule engine, automatic sensitive data masking, SQL risk prevention, Spec pre-checks, full-link auditing, and emergency circuit-breaking. Enterprise-grade security is enabled by default with no complex configuration required, allowing enterprises to confidently deploy agents for production data. |
Agent Ready | Agent-oriented data services and Agent services extend AI Agent capabilities through premium DataSkill and MCP integrations, enabling deep collaboration between data and agents. |
Simplicity and ease of use | A one-stop intelligent data development platform with ready-to-use Data Agent products, complemented by a rich set of skills and low-code development tools to accelerate AI business deployment. |
Billing
AI-Native Database Service offers multiple billing options, including free trials, subscription, and pay-as-you-go. You can purchase one or more individual service modules based on your requirements.
Module | Feature | Billing method |
Data Agent | Analytics Agent | Subscription |
Forecast Agent | Free during the invitational preview | |
DataBridge Agent | Free during the invitational preview. Subscription and pay-as-you-go are supported after general availability. | |
DAS Agent | Subscription, pay-as-you-go | |
Agent Data Gateway | Free during the invitational preview | |
Enterprise-grade database Agent Developer Platform | Dify | Subscription, pay-as-you-go |
Supabase | Subscription, pay-as-you-go | |
Managed Data Agent | Free during the invitational preview | |
Proactive Agent | Free during the invitational preview. Subscription and pay-as-you-go are supported after general availability. | |
AI Data Preparation | Multimodal Data Preparation | Pay-as-you-go |
AI asset and enterprise knowledge services | Meta Agent | Subscription |