Appian Task Management Chatbot using Ollama and Postgres - n8n Workflow

Automation Specialists: Seeking complex, tool-based n8n templates for enterprise systems.
Developers: Interested in integrating local AI (Ollama) with business process management systems like Appian.
IT Administrators: Needing an extensible n8n workflow solution for dynamic task creation and retrieval via conversational input.
n8n Power Users: Who want to see how to use the AI Agent n8n node with custom tool definitions and persistent memory.

This comprehensive n8n workflow creates a powerful conversational AI assistant dedicated to managing tasks within Appian. Unlike solutions relying solely on external services, this template utilizes the lmChatOllama n8n node, allowing you to run the Large Language Model (Qwen 2.5) locally, enhancing privacy and control.

When a message is received (via the n8n trigger or webhook), the input is routed to the AI Agent. This n8n node is configured with explicit tools—HTTP Request n8n node definitions—that interact with the Appian API (list tasks, create tasks, list task types). The Postgres Chat Memory n8n node ensures context is maintained across sessions. This robust n8n workflow provides a production-ready template for internal chatbots managing critical business processes.

This n8n workflow starts with one of two possible n8n trigger mechanisms: the When chat message received node or the standard Webhook n8n trigger.

1. Input Normalization: The Normalize Chat Input n8n node standardizes the incoming message data, extracting the user's text, conversation ID (for session memory), and username. These variables are crucial for personalizing Appian API calls.


2. AI Agent Activation: The centralized AI Agent n8n node receives the normalized input. It uses the Ollama Chat Model (Qwen 2.5) as its brain and the Postgres Chat Memory n8n node to recall past conversations.


3. Tool Selection: Based on the user's request (e.g., "List my open tasks" or "Create a new task titled 'Review Q4 Report'"), the AI Agent determines which specific Appian-related n8n node tool to execute.


4. Appian Interaction: The chosen HTTP Request n8n node tool executes the API call to Appian. Crucially, complex parameters (like task title or description for task creation) are dynamically extracted by the AI Agent using the $fromAI function within the n8n node configurations.


5. Response Generation: The AI Agent receives the API response, synthesizes a human-readable reply, and passes it to the Prepare Response n8n node.


6. Final Output: The Respond to Webhook n8n node sends the agent's output back to the originating chat platform, completing the interactive cycle of this powerful n8n workflow.

To deploy this advanced n8n workflow, follow these steps:

1. Import: Copy the provided JSON and import it as a new n8n workflow.


2. Ollama Setup: Ensure you have Ollama running locally and the qwen2.5:7b model pulled, as referenced by the Ollama Chat Model n8n node.


3. Postgres Setup: Configure your Postgres database credentials and ensure the necessary schema for LangChain memory is initialized.


4. Appian Credentials: Set up a new generic credential (OAuth2 API or appropriate type) for accessing the Appian Web API, linking it to the List Tasks (Appian), List Task Types (Appian), and Create Task (Appian) n8n node tools.


5. Template Variables: Update the Template Vars n8n node with your specific Appian base URL (baseUrl).


6. Webhook Configuration: If using the Webhook n8n trigger, update the path and ensure your external chat application is configured to send POST requests to this n8n endpoint.

This n8n workflow leverages several specialized and core n8n node types:

Webhook / When chat message received (n8n trigger): Either of these serves as the entry point for the n8n workflow, handling incoming conversation data.
Normalize Chat Input (Set n8n node): Essential for mapping varying input formats (from different n8n triggers) into standardized variables (chatInput, sessionId, username).
Ollama Chat Model (n8n node): Defines the local LLM running via Ollama (qwen2.5:7b), providing the intelligence layer for the n8n workflow.
Postgres Chat Memory (n8n node): Provides long-term conversational memory, enabling the AI Agent to maintain context across multiple messages.
AI Agent (n8n node): The core decision-making hub. It uses the system message: "You are a helpful assistant that can list and create Appian tasks..." and routes tasks to the appropriate tools.
List Tasks (Appian) / Create Task (Appian) (HTTP Request Tools n8n node): These are customized HTTP Request n8n nodes configured as 'Tools' for the AI Agent. They use the $fromAI expression (e.g., for startIndex, title) to dynamically pull required data extracted by the LLM from the user's prompt.