đź’¬ Build an IT support assistant chatbot leveraging existing support portal
⚡ 6,187 views · 💬 Support Chatbots
Description
This n8n template demonstrates how you can leverage existing support site search to power your Support Chatbots and agents.
Building a support chatbot need not be complicated! If building and indexing vector stores or duplicating data isn’t necessarily your thing, an alternative implementation of the RAG approach is to leverage existing knowledge-bases such as support portals. In this way, document management and maintenance of your support agent is significantly reduced.
Disclaimer: This template example uses AcuityScheduling’s help center website but is not associated, supported nor endorsed by the company.
How it works
- A simple AI agent is connected with chat trigger to receive user queries.
- The AI agent is instructed to fetch information from the knowledge-base via the attached custom workflow tool (aka “knowledgebase tool”).
- There is no step to replicate the entire support articles database into a vector store. You may choose not too because of time, cost and maintainence involved.
- Instead, the tool leverages the existing support portal’s search API to retrieve knowledge-base articles.
- Finally, the search results are formatted before sending an aggregated response back to the agent.
How to use?
- Customise the subworkflow to work with your own support portal API and format accordingly.
- Try the following queries
- How do I connect my icloud to acuityScheduling?
- How do I download past invoices for my Acuity account?
Requirements
- OpenAI for LLM.
- If your organisation’s APIs require authorisation, you may need to add custom credentials as necessary.
Customising this workflow
- Add additional tools to reach other parts of your internal knowledgebase.
- Not using OpenAI? Feel free to swap but ensure the LLM has tools/function calling support.
đź”— Nodes Used
HTTP Request, Execute Workflow Trigger, AI Agent, OpenAI Chat Model, Simple Memory, Call n8n Workflow Tool
đź“Ą Import
Download workflow.json and import into n8n:
Workflow menu → Import from File