Open Editions

Open Editions

Come for answers. Stay for best practices. All we’re missing is you.

 View Only

Introducing AI Agent Task: Advanced AI Integration for BAMOE Business Processes

By Adarsh V K posted 2 days ago

  

Starting with BAMOE v9.3.1, we are excited to introduce the AI Agent Task. This is a new BPMN workflow node that helps you use any AI agent or flow you build in Langflow, right inside your BPMN workflows. Langflow is an open-source tool where you can create AI logic using a simple drag-and-drop editor. With this feature, you can connect your Langflow agents or flows to your business processes easily, making your workflows smarter and more flexible. This integration is simple to set up and does not need any extra code or complicated steps.

What is Langflow?

Langflow is an open-source tool for building and running AI workflows. It provides a drag-and-drop interface where you can create, test, and manage AI agents or flows without writing much code. Langflow supports many popular AI models, including OpenAI, watsonx, Anthropic, and Ollama. It also works with vector databases and a wide range of AI tools. You can connect to different data sources, models, and vector stores, or add your own custom components. Langflow is designed to make it easier to build, test, and use AI in your business, whether you are working on simple or complex workflows.
Langflow lets you:
  • Design and deploy AI agents and flows visually
  • Swap and compare models and components with ease
  • Run, share, and collaborate on flows
  • Deploy on your own infrastructure or use a secure, enterprise-grade cloud
  • Move from notebook to production in minutes

Why did we choose Langflow?

We wanted to give BAMOE users limitless control and flexibility for building intelligent automation. Langflow's visual state flows, reusable components, and rapid iteration empower both business and technical users to create and adapt AI-powered processes. You can focus on creativity and business value, not on wiring up APIs or writing glue code.
Langflow is open-source (MIT License), actively developed, and trusted by thousands of developers. Whether you want to run a single agent or a fleet, Langflow gives you the tools to scale, experiment, and innovate.
Learn more about Langflow:


What is the AI Agent Task?

The AI Agent Task is a custom node you can add to your BPMN model. It connects to your own Langflow instance, where you visually build AI flows using pre-built components and connect to popular AI models like OpenAI, watsonx, and Ollama. Once connected, you can select any of your Langflow agents or flows, configure their inputs, and test them right in the property panel.

Design-time

From Canvas or Developer Tools for VS Code, simply connect your Langflow account by providing the service URL and API key. After connecting, you can drag the AI Agent Task onto your workflow, pick an agent or flow from your Langflow instance, and map process variables as inputs and outputs. The property panel offers an advanced input editor with variable auto-complete and a preview table for testing with sample data. You can run test executions and see real results before deploying.

Step-by-step setup in Canvas:
  1. Click the profile icon in the top right corner of Canvas.
  2. Select "Connected accounts" from the menu.
  3. Choose Langflow from the list of AI providers.
  4. Enter a display name, your Langflow Service URL, and API Key.
  5. Click "Connect" to save your account.

Step-by-step setup in Developer Tools for VS Code:
  1. Click the profile menu in the bottom left corner of VS Code.
  2. Select Langflow as your AI provider.
  3. Enter your Langflow Service URL and API Key.

Configuring the AI Agent Task in the BPMN Editor:
  1. Drag and drop the AI Agent Task node onto your BPMN diagram.
  2. In the property panel, select your connected Langflow account.
  3. Choose the agent or flow you want to use from the dropdown, or enter the Agent ID directly.

  4. View details about the selected agent, such as its ID, description, tags and MCP enabled.

  5. Provide input for the agent. You can use the default aiAgentInput variable, and include process variables using {{variable-name}} format.
  6. Use "Advance Mode" in the input field for syntax highlighting and auto-suggestions for variables (press Ctrl+Space or Cmd+Space).
  7. Set up input and output data mapping as needed. By default, aiAgentInput and aiAgentOutput are used, but you can add more variables.
  8. Use the Preview Table to test your input variables with temporary values.
  9. Click "Test Agent" to run the agent and see the output in real time. Adjust your input as needed until you get the desired result.


Runtime Integration

To use the AI Agent Task in production, add the appropriate work item handler dependency (for Quarkus or Spring Boot) to your project and configure your Langflow credentials in `application.properties`. At runtime, the handler will execute the selected Langflow flow, passing in your mapped variables and returning results to your process variable.

Step-by-step runtime setup:
1. Add the correct work item handler (WIH) dependency to your project:
- For Quarkus: bamoe-ai-agent-task-work-item-handler-quarkus
- For Spring Boot: bamoe-ai-agent-task-work-item-handler-spring-boot
2. In your `application.properties` file, set the following properties with your Langflow details:
- bamoe.workflow.ai-agent-task.provider.langflow.base-url=<your-langflow-host>
- bamoe.workflow.ai-agent-task.provider.langflow.api-key=<api-key>
3. Deploy your business service as usual. When the workflow reaches the AI Agent Task node, the handler will connect to Langflow, send the input, and return the output to your process variables automatically.

 

Why we built it this way ?

We wanted to give you the power to build and reuse advanced AI logic—multi-step agents, tool use, and stateful flows—without leaving your BPMN editor. By leveraging Langflow, you get a rich ecosystem of AI integrations and a visual editor, while BAMOE handles the orchestration and data mapping. The result: faster iteration, more flexible automation, and a seamless experience for both business and technical users.

 

0 comments
27 views

Permalink

https://community.ibm.com/community/user/blogs/adarsh-v-k/2025/11/28/introducing-ai-agent-task-advanced-ai-integration