DataBridge Agent is a multi-source data collection and parsing tool provided by AI-Native Database Service. It ingests data from databases, documents, web pages, and other sources, and converts the data into standardized structured formats for AI model training, RAG workflows, and data analytics.
What is DataBridge Agent
When training large models or performing data analysis, enterprises often need to integrate data from databases, web pages, documents, and other sources. However, complex data formats, inconsistent quality, and the lack of a unified collection tool make data ingestion and processing inefficient.
DataBridge Agent encapsulates complex data processing workflows into an independent agent based on the core data collection and parsing capabilities of Alibaba Cloud database tools, enabling unified access to multiple data sources. You can convert raw data into standard formats such as JSON, CSV, and Parquet through simple configuration, serving multiple downstream systems and bridging disparate data links within your enterprise.
Core capabilities
DataBridge Agent provides the following three core capabilities:
Multimodal data preparation and retrieval: Supports data collection from databases, unstructured documents, web pages, and other data sources. Built-in OCR and intelligent parsing engines automatically identify and extract fields, headers, and hierarchical relationships, and convert data into standardized structured formats.
Data exploration and analysis: Automatically analyzes and evaluates the quality of collected data, helping you quickly understand data distribution and detect anomalies. Delivers validated, high-quality data for downstream AI applications.
Pipeline operations and O&M: Provides visual task management and monitoring capabilities through the console. Supports task scheduling, running status monitoring, and exception alerts to keep data collection pipelines running reliably.
Supported data sources
DataBridge Agent supports the following types of data sources:
Type | Scope |
Databases | Supports mainstream relational and analytical databases, such as:
|
Unstructured documents | PDF, Word, Excel, PPT, Markdown, and more. Includes built-in OCR to parse text and tables from images or scanned files. |
Web content | Accurately captures web data by extracting HTML page structures or simulating API requests. |
Output formats
DataBridge Agent converts collected raw data into the following standard formats, which can be directly used for large model training or data analytics:
JSON: Suitable for API integration and flexible data exchange scenarios.
CSV: Suitable for tabular data analysis and batch import scenarios.
Parquet: Suitable for big data analytics and efficient columnar storage scenarios.
Use cases
Scenario | Description |
Data preparation for large model training | Quickly collect and structure massive, multi-source data to provide high-quality training corpora for large language models (LLMs). |
Data input for agent workflows | Provides precise, real-time external data for applications such as Retrieval-Augmented Generation (RAG) and process agents, improving the accuracy and timeliness of AI applications. |
Cross-cloud and hybrid cloud data integration | Centrally extract and integrate data from multiple systems, including on-premises data centers, private clouds, and other public clouds. |
Automated document processing | Batch parse business documents in formats such as PDF and Excel into structured data for BI analytics, report generation, or data archiving. |
Web information gathering | Crawl and structure content from e-commerce, news, and public opinion websites for market analysis, semantic understanding, and knowledge graph construction. |
Billing
DataBridge Agent uses a subscription billing model with predefined specifications. Specifications are divided into Basic and Enterprise editions. You can choose the edition that best fits your data volume, connector scope, and processing mode.
All Alibaba Cloud accounts that have completed identity verification are eligible for a one-time free trial of the Small specification under the subscription model for one month.
Pricing tiers
|
Category |
Basic Edition |
Enterprise Edition |
|||
|
Tier |
small |
medium |
large |
standard |
pro |
|
Monthly fee (USD) |
275 |
550 |
1,375 |
14,000 |
70,000 |
|
Connector type |
Basic only |
Basic only |
Basic and advanced |
Basic and advanced |
Basic and advanced |
|
Connector limit |
1 |
3 |
3 |
10 |
30 |
|
Token limit |
300 million |
700 million |
2 billion |
24 billion |
120 billion |
|
Processing mode |
Fixed pipeline |
Fixed pipeline |
Fixed pipeline |
AI Mission dynamic orchestration |
AI Mission dynamic orchestration |
|
Advanced link development assistant |
— |
— |
— |
Optional add-on (USD 7,000/month or USD 14,000/month) |
Optional add-on (USD 7,000/month or USD 14,000/month) |
|
Target customer |
Proof of Concept (PoC) and solution evaluation |
Startups and micro-enterprises |
Mid-sized enterprises |
Large enterprises |
Multinational groups and leading enterprises |
Connector type description
|
Type |
Protocol characteristics |
Typical data sources |
|
Basic |
Standard protocols (S3 API, SMB or NFS, JDBC). Configure your AccessKey once and use it directly. |
Object Storage Service (OSS), Amazon S3, Azure Blob, File Storage NAS, FTP, MySQL, PostgreSQL, web pages, and more. |
|
Advanced |
Private APIs, OAuth 2.0, and multi-level permissions. Requires ongoing maintenance. |
Lark, DingTalk, WeCom, Confluence, SharePoint, Salesforce, Outlook, Gmail, and more. |
Access DataBridge Agent
You can access DataBridge Agent by performing the following steps:
Log on to the AI-Native Database Service console.
In the left-side navigation pane, click Data Ingestion.