When training large models or performing data analysis, enterprises often need to integrate data from various sources such as databases, web pages, and documents. However, complex data formats, inconsistent quality, and the lack of a unified collection tool lead to inefficient data ingestion and processing. DataBridge Agent, from Alibaba Cloud Data Transmission Service (DTS), addresses these challenges. It is a multi-source data collection and parsing tool that efficiently ingests, parses, and transforms heterogeneous data from different sources into a unified structured format, and provides high-quality input data for downstream applications like AI model training and data analytics.
DataBridge Agent will be commercially available starting 00:00:00 (Singapore Time) on August 1, 2026 and charged based on the tier you select. For pricing details, see Billing.
Overview
DataBridge Agent is a multi-source data collection and parsing tool from Alibaba Cloud Data Transmission Service (DTS). It integrates the core data collection and parsing capabilities of DTS for databases, web pages, and documents with its robust data transmission and intelligent O&M capabilities. The agent helps you efficiently acquire and standardize heterogeneous data on a single, unified platform.
The tool encapsulates complex data processing workflows into an independent agent. This enables unified access to multiple data sources, one-time parsing into a standard format, and serving multiple downstream systems, thereby connecting disparate data links within your enterprise.
Benefits
Wide data source support
Eliminate the need to develop separate adaptation logic for different data sources. Connect to all of them through a unified agent, significantly reducing development and maintenance costs.
|
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. |
Powerful automated parsing and structuring
Built-in data parsing engines automatically identify and extract fields, headers, and hierarchical relationships from your data. You can also define custom rules to map data for specific business needs. You can convert raw data into common structured formats like JSON, CSV, or Parquet with a single click. The output is ready for direct use in large model training or data analysis.
Seamless integration with the AI ecosystem
-
Acts as a data preprocessing tool for large models, providing clean and consistently formatted training data.
-
Adapts to various agent workflows, such as Retrieval-Augmented Generation (RAG), providing them with real-time, accurate external data.
-
Provides standard API calls for easy integration into your existing AI systems or automation platforms.
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 like Retrieval-Augmented Generation (RAG) and process agents, which improves 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 like 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. |
|
Data governance and cleaning |
Serves as an upstream step in ETL processes to standardize raw data. This improves the quality and consistency of data loaded into your data stores. |
Billing
DataBridge Agent uses tiered pricing with a subscription (monthly or yearly prepaid) billing method. The tiers are grouped into two editions: Basic Edition and Enterprise Edition. You can choose a suitable tier based on your data volume, connector usage, and processing mode.
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. |