Overview
As IoT deployments grow more sophisticated and data streams become richer and more complex, intelligent automation and real-time decision-making are no longer optional — they’re essential. That’s why we’re thrilled to introduce the AI-Agent Manager, a powerful new addition to the Cumulocity IoT platform. It empowers you to create, configure, and centrally manage AI-powered assistants, seamlessly integrating them into your workflows.
We showcased a live demo of the AI-Agent Manager in our What’s New webinar in July — watch the recording here:
In this article, we’ll explore how the AI-Agent Manager transforms your Cumulocity IoT deployment into an intelligent AIoT ecosystem, harnessing your device data and extending its reach into external systems via the Model Context Protocol (MCP).
Revolutionizing IoT Intelligence with No-Code AI
The AI-Agent Manager represents a significant leap forward in democratizing AI capabilities within IoT environments. By providing an intuitive interface for agent creation, it eliminates the traditional barriers between IoT data and artificial intelligence, allowing you to harness the power of large language models directly within your tenant environment.
Key Benefits at a Glance
1. Rapid, No-Code Agent Creation
Configure AI assistants through an intuitive UI interface. Select from industry-leading large language models including GPT-4o, Claude, and Gemini, then define each agent’s behavior, tool access, and API integrations without any coding requirements. This approach significantly reduces the technical expertise needed to implement AI solutions in your IoT environment.
2. Centralized Agent Management
Manage multiple agents from a single dashboard, streamlining the entire agent lifecycle. Update system prompts, adjust tool configurations, and coordinate agent workflows across your organization with ease. This centralized approach ensures consistency and simplifies maintenance across complex IoT deployments.
3. Deep Cumulocity Integration
Agents seamlessly integrate with core Cumulocity capabilities, including inventory management, events, measurements, and alarms. This native integration eliminates the need for additional setup or data mapping, allowing agents to access real-time device context and tenant-specific data instantly.
4. Extensible via MCP Servers
The Model Context Protocol (MCP) server support provides plug-and-play connectivity to external systems and custom tools. Whether you need to query third-party APIs, invoke cloud functions, or trigger downstream workflows, MCP integration makes it simple to extend your agents beyond the Cumulocity ecosystem.
Would you like to set up your own Cumulocity MCP server ? : Read More here: Supercharging IoT with Agentic Systems: The Model Context Protocol (MCP)
5. Sandbox Testing & Validation
Preview and validate agent behaviors in a safe, isolated environment before production deployment. This testing capability ensures that your agents perform as expected and helps prevent unintended consequences in live IoT environments.
Core Configuration Workflow
The AI-Agent Manager follows a streamlined five-step workflow that guides you from concept to production deployment:
Step 1: Select Your LLM Model
Choose the optimal large language model for your specific use case. Consider GPT-4o for conversational tasks or Claude for large-scale data analysis. Each model brings unique strengths to different IoT scenarios.
Important : Bring Your Own Key : The API Key in the AI Agent Manager allows you to connect and use your existing LLM provider for your specific use case, giving you full control over your enterprise and asset data.
Step 2: Define Agent Behavior & Tools
Configure system prompts that define how your agent should interact with users and respond to different scenarios.
Configure Advanced Settings: Advanced agent settings (JSON format), e.g., for Vercel AI SDK tool use. This merges with the system prompt settings.
Step 3: Configure Tools
Grant agents access to built-in plugins, REST APIs, or custom MCP tools based on your specific requirements.
Step 4: Supply Dynamic Context
At runtime, inject relevant managed object metadata, device information, or tenant-specific data to ensure agents operate with the most current and relevant information available.
Step 5: Test in Sandbox
Simulate real-world interactions within the controlled test environment. Review agent decisions, validate tool calls, and ensure response flows meet your expectations before moving to production.
Step 6: Deploy & Monitor
Deploy validated agents into your production environment and continuously monitor their performance. Track usage metrics, tool invocations, and conversation flows to optimize agent effectiveness over time.
Practical Use Cases
Predictive Maintenance Assistant
Create an AI agent that monitors device measurements and correlates them with historical maintenance data. The agent can proactively identify potential equipment failures and recommend maintenance actions based on real-time sensor data and external weather APIs accessed through MCP servers.
Virtual Plant Operations Assistant
Deploy an intelligent assistant that provides plant operators with instant insights by analyzing device events and alarm patterns. The agent can suggest optimal operational parameters and provide context-aware troubleshooting guidance.
Supply Chain Optimization Agent
Build an agent that combines IoT sensor data from your device inventory with external logistics APIs to optimize supply chain operations and provide intelligent recommendations for inventory management.
Business Impact and Strategic Value
The AI-Agent Manager delivers significant business value across multiple dimensions:
- Accelerated AIoT Adoption: Enable rapid deployment of generative AI capabilities on existing IoT infrastructure, reducing time-to-market from months to hours
- Enhanced Co-Innovation: Facilitate collaborative development of novel AI use cases directly within the Cumulocity platform
- Reduced Development Overhead: Eliminate custom coding requirements through drag-and-drop configuration and out-of-the-box integrations
- Strategic Positioning: Strengthen your organization’s position as an AIoT leader by leveraging cutting-edge AI capabilities
Accessing Private Preview
We’re inviting a select group of customers to join the AI-Agent Manager Private Preview - a unique opportunity to shape the future of AI-powered IoT. As a participant, you will:
- Gain early access to agent creation capabilities and MCP tool integration
- Collaborate directly with our engineering and product teams to explore innovative AIoT scenarios
- Contribute feedback that drives iterative development and influences the product roadmap
- Receive regular updates on new features and enhancements
Important: General availability will be determined based on insights gathered during this preview phase.
How to Get Started
Contact your Customer Success Manager (CSM) to request early access, share your use cases, and participate in validation activities to ensure AI capabilities align with your operational goals.