How to Set Up an n8n Workflow to Connect Microsoft Teams with an Ollama Agent?

Below is one approach to build a workflow where Microsoft Teams messages trigger an Ollama agent response using n8n:

1. Listen for User Messages in Microsoft Teams

• Use the Microsoft Teams node configured as a trigger or polling node.
• For example, you can use an operation that gets messages from a specific channel or chat.
• Follow the instructions in the Microsoft Teams node documentation to set up credentials and select the correct operation (such as “Channel Message Create” to send messages or “Get Many” to check for new ones).

2. Extract and Process the User’s Message

• Once a new message is received, use a Function node or set values with expressions to extract the text that the user submitted.
• Keep in mind that if your node is a sub-node (like the Ollama Chat Model node), expressions only reference the first item. That means you need to ensure your workflow properly handles or routes the incoming data.

3. Send the Message to the Ollama Chat Model Node

• Add the Ollama Chat Model node to your workflow.
• Configure it by setting the base URL (for example, http://127.0.0.1:11434 if encountering IPv6 issues or http://localhost:11434 when possible) and any other required parameters.
• The node will use the incoming message as input and generate the agent’s response.
• Ensure that your local Ollama instance is correctly running and accessible from your n8n (Docker) container.

4. Return the Response Back to Microsoft Teams

• Once the Ollama Chat Model node outputs the response, add another Microsoft Teams node that sends a message back to the user or the channel.
• Configure this node with the appropriate operation (e.g., “Create Chat Message”) to post the generated response.
• Use expressions to map the output from the Ollama node to the message body.

5. Testing and Troubleshooting

• Verify proper credentials for both Microsoft Teams and your Ollama local instance.
• If working with Docker, ensure that network configurations allow n8n and Ollama to communicate—especially if either is running in separate containers.
• For example, if only n8n is in Docker, remember to configure the host address in your Ollama credentials to use host.docker.internal:11434.

Example Workflow Outline

MERMAIDCopied
flowchart TD
    A[Microsoft Teams: New Message Trigger] --> B[Extract Message Data]
    B --> C[Ollama Chat Model: Generate Response]
    C --> D[Microsoft Teams: Post Response Message]

Final Notes

• Make sure you follow the Ollama credentials instructions for connecting to the locally hosted Ollama instance.
• When integrating multiple nodes, review each node’s nuances—especially with sub-nodes processing single items from multiple inputs.
• Adjust your workflow based on your specific Microsoft Teams events and the type of conversation you want to automate.

This setup will allow your n8n workflow to “listen” for Microsoft Teams messages and then channel those messages to an Ollama agent, which generates a reply that is then sent back to Teams.