SHANGHAI, July 21, 2026 /PRNewswire/ — Tencent Cloud, the cloud business of global technology company Tencent, today announced a major enhancement to its Agent Development Platform (ADP) International Edition, evolving it into a comprehensive, enterprise-grade AgentOps platform designed to help enterprises move AI agents from experimentation into production at scale. The enhancement was unveiled at the 2026 World Artificial Intelligence Conference (WAIC) in Shanghai and is now available to customers worldwide.
Delivered in ADP’s 4.0 international release, the update introduces two headline capabilities — the Smart Desk and Claw Mode, powered by the Agentic Loop. Together with Connector, Skills, Knowledge Base, MCP and Agent Portal, ADP 4.0 now connects the entire enterprise agent lifecycle — from building and connecting, to distribution and governance — giving enterprises a single foundation to develop, deploy and manage AI agents reliably and at scale.
Today, enterprises seeking to rebuild their operations with AI Agents often face three critical challenges:
- Individual pilots are easy to initiate but difficult to scale;
- Powerful models are challenging to integrate with existing legacy systems;
- Deployed Agents present difficulties in security governance and cost control.
Tencent Cloud ADP 4.0 addresses these challenges systematically by unlocking long-horizon, self-coding capabilities for complex tasks through Claw Mode, connecting enterprise systems and capability assets through nearly 40 Connectors and over 150 Skills, while moving security governance upstream to the very source of development. This enables enterprises to rapidly develop Agents while keeping them running in real business environments in a stable, secure, and controllable way.
To date, Tencent Cloud ADP has been deployed across retail, finance, government, manufacturing, and other industries, where teams have used AI Agents to support on tasks such as intelligent Q&A, customer service, marketing content generation and business analysis.
How Smart Desk Serves as a Unified Agent Entry Point
The newly launched Smart Desk serves as the unified, user-facing entry point of Tencent Cloud ADP 4.0, addressing two needs at once: enterprises can build Agents at scale, while employees can complete everyday tasks with a single sentence. Within the same workspace, users can build scalable Agents for their organization using natural language, or handle personal tasks just as they would with an everyday office assistant. Agent capabilities can also be integrated into existing enterprise systems—office software, ERP, CRM, customer service, and more—so that Agents “grow” on top of existing workflows without tearing anything down.
Smart Desk dramatically cuts the cost of building. Setting up a knowledge-based intelligent Q&A assistant once required manually configuring more than a dozen parameters. Now a single sentence can complete the entire configuration automatically, enabling it to go live within minutes. For more complex workflows such as multi-intent customer service, users simply describe the logic, and the system automatically generates a complete flow—intent routing, knowledge-base Q&A, human handoff, and more—packaged and ready to import. When a single Agent isn’t enough, users can orchestrate a collaborative team of Agents with a single command, with the system automatically assigning responsibilities, prompts, and task handoffs. The Smart Desk also integrates with the Skills Marketplace, packaging mature capabilities—such as contract review, financial analysis, and invoice entry—into reusable enterprise-grade Skills that can be shared and continuously refined across the organization. In doing so, Smart Desk brings “usable by everyone” and “enterprise-grade control” together in a single entry point.
How Upgraded Claw Mode Resolves Complex, Long-Horizon Tasks
For complex tasks that lack fixed rules, span long chains, and resist form-based configuration, Tencent Cloud Agent Development Platform (ADP) 4.0 adds a fourth building mode—Claw Mode. It can create Agents that autonomously write and run code in a cloud sandbox, call enterprise Skills, and execute long-running tasks. Once created, these Agents can be integrated into business systems via API/Web interfaces and reach employees and customers through channels such as LINE and Telegram. Creators do not need to fill out any forms; they simply describe their needs in natural language, and the platform automatically generates prompts, mounts knowledge bases, configures tools, and orchestrates workflows.
ADP 4.0 also supports two-way invocation between Agents and Workflows: Workflows handle steps with clear-cut rules, balancing stability with predictable token consumption. When unstructured judgment or open-ended tasks arise, they invoke a Claw Mode Agent to fully leverage the model’s capabilities. This gives enterprises the flexibility to integrate deterministic processes with intelligent decision-making.
Connecting Enterprise Systems, Knowledge, and Tools to Embed Agents into Workflows
To address the pain point of “powerful models that cannot reach enterprise production systems,” ADP 4.0 uses Connector, Skills, plugins, Knowledge Base, and MCP to turn an enterprise’s scattered business resources into callable and reusable AI assets.
The Connector framework offers nearly 40 curated Connectors in the first batch, supporting integration with high-frequency systems such as CRM, ERP, ticketing, customer service, knowledge bases, and document systems, and covering mainstream international SaaS applications like Google Workspace, Confluence, and Jira. Agents can directly read, retrieve, and perform operations such as querying, analysis, generation, and routing.
The upgraded Skills Marketplace now supports 150+ Skills. Business teams can package custom Skills into shared enterprise plugins that enter the enterprise zone for use only after passing security checks and approval.
More than 50 scenario-based templates and industry-curated applications shorten the path from zero to one. The upgraded Model Marketplace enables enterprises to select leading international models alongside their own self-hosted models, with native compatibility for the OpenAI standard protocol.
In addition, ADP 4.0 supports multiple entry points—platform, browser, Office, Chat channels, and API/SDK—embedding Agents into employees’ daily work scenarios.
End-to-End Governance from Development to Launch: Clearing the “Security Gate” Before Production
To tackle the challenge of “Agents that run but whose security and cost cannot be controlled,” ADP 4.0 moves governance upstream to the source of development, building a security control system that spans the full Agent lifecycle around permission management, Skills governance, runtime observability, and deployment compliance.
For permission management, ADP 4.0 supports a layered architecture at the enterprise, workspace, and application levels, combined with a Role-Based Access Control (RBAC) role-permission matrix to achieve dual-dimension isolation of functional and data permissions. For Skills governance, custom Skills must pass security checks—static code scanning, data-access review, outbound-network review, dependency allowlisting—and multi-level approval before entering the enterprise zone to become shareable and schedulable enterprise assets. For observability and governance, enterprises can use Agent Portal to centrally manage Agents across platforms and scenarios, and use business and resource dashboards to monitor key metrics such as call volume, response quality, operating cost, and errors in real time, enabling rapid issue identification and optimization.
Wu Yunsheng, Vice President of Tencent Cloud and Head of the Tencent Cloud Agent Development Platform, emphasized that enterprise-grade Agents are not merely about who can build them the fastest, but rather about who can keep Agents operating on the front lines of business in a stable, secure, and sustainable manner. ADP aims to provide enterprises with a production-ready AgentOps foundation—one that enables them to build Agents quickly, connect these Agents to systems and distribute them to the front lines, and operate them continuously under a framework of permissions, security, evaluation, and observability, truly unlocking productivity value.
Learn more about Tencent Cloud ADP: Build AI Agents with RAG+LLM, Workflow & Multi-Agent Capabilities
###
About Tencent Cloud:
Tencent Cloud, one of the world’s leading cloud companies, is committed to creating innovative solutions to resolve real-world issues and enabling digital transformation for smart industries. Through our extensive global infrastructure, Tencent Cloud provides businesses across the globe with stable and secure industry-leading cloud products and services, leveraging technological advancements such as cloud computing, Big Data analytics, AI, IoT, and cybersecurity. It is our constant mission to meet the needs of industries across the board, including the fields of gaming, media and entertainment, finance, healthcare, real estate, retail, travel, and transportation.
