Add preferred source
UiPath has added AI development, data connectivity, and security management features to its enterprise automation platform and officially launched the Coding Agent it first unveiled in May. The Coding Agent generates automation programs from natural-language requests and integrates with external AI development tools including Anthropic’s Claude Code, OpenAI’s Codex, Cursor, and Google Antigravity. For repetitive tasks, the AI agent “Delegate” is deployed, though final judgment and approval remain with humans. The company strengthened execution controls such as the Runtime Checker and Model Hub, and expanded support for on-premises operations in Linux environments.
Key Elements

Enterprise automation platform company UiPath is putting front and center a capability for AI agents to directly access internal systems and handle work. The hallmark is that while more tasks are handed to AI, management controls are attached so enterprises can govern data access and execution.
UiPath said on the 16th that it has added AI development, data connectivity, and security management features to its automation platform and improved existing functions. The Coding Agent unveiled last May is now being offered as an official version. The company said it squarely targets the security and permission problems enterprises hit when putting AI agents into actual work.
The most notable change is the openness of development tools. The officially launched Coding Agent helps developers generate automation programs or modify existing work procedures when they make requests in natural language. Beyond that, external AI development tools such as Anthropic’s Claude Code, OpenAI’s Codex, Cursor, and Google Antigravity can also be integrated and used on the platform. These tools read task histories, execution records, and connection information registered on the UiPath platform and use them for automation development.
For repetitive tasks, the enterprise AI agent “Delegate” is deployed. Delegate supports gathering information from multiple systems, modifying records, and attaching supporting documents. It also has a function to save tasks that staff repeat and re-execute them when needed. However, the structure keeps major task judgments and final approval with humans.
The data connectivity segment was also overhauled. Through the database hub, users can access Microsoft SQL Server, Oracle, Databricks, and others without writing separate code. AI agents running in the cloud also connect to internal serverstial management systems as-is. Support for the Model Context Protocol (MCP), which attaches external AI tools, is also included
The data management feature “Data Fabric” has been improved so that AI agents can use data stored in SAP or Salesforce without separately copying it. The explanation is that enterprises can connect information scattered across multiple systems to automation work while maintaining existing access permissions.
As AI’s access scope has widened, control functions have also been strengthened. The “Runtime Checker” checks in real time whether AI agents comply with security rules set by the enterprise while performing work. In the “Model Hub,” users can check which AI model is being used and set access permissions by model or tool. Functions are also provided to evaluate AI-generated answers against enterprise standards or inspect compliance. UiPath has integrated its management system so that the same security policy applies to AI agents, existing automation programs, and employees alike.
Demand for running AI on in-house servers was also reflected. The support scope was expanded so that the full range of automation platform functions can run in Linux environments. Financial companies and public institutions with restrictions on using external clouds can now operate AI agents and automation systems on internal infrastructure.
Raghu Malpani, UiPath’s Chief Product and Technology Officer (CPTO), said, “The UiPath platform provides a map so customers can set the right direction for their agentic transformation at the speed and context their individual enterprises need,” adding, “It gives confidence that all of this stays within the boundaries set by the business.”
The announcement aligns with the trend in which “who controls AI’s actions” has emerged as a key issue in the enterprise automation market as AI agent adoption takes off. The moment AI connects to internal databases and work systems, the importance of permission management and audit trails grows, and UiPath aims to differentiate by managing both its own development tools and external AI tools on a single platform.
The market takes note of the fact that it actively embraces external development tools. Claude Code, Codex, Cursor, and Antigravity are tools being rapidly adopted in software development settings recently. UiPath’s decision to allow them to access its platform’s execution records and connection information reads as a strategy to turn AI tools used by individual developers into enterprise-level automation assets.
For enterprise customers, the key question is whether this can reduce the security review burden that has been cited as an obstacle to AI agent adoption. As the financial and public sectors have faced significant restrictions on using external clouds, Linux-based on-premises operation support is expected to determine the actual pace of adoption. UiPath did not disclose specific pricing policies or launch schedules in this announcement.
Once added, BigGo Finance appears first in Google Search Top Stories, so you get the broadest, most up-to-the-minute, and most comprehensive global financial news first.
