📖 Answer procurement policy questions with Google Drive, Gemini, and webhooks
⚡ 15 views · 📖 Internal Wiki & Knowledge Base
Description
Quick overview
This workflow exposes a webhook-based helpbot that answers employee questions using a policy PDF from Google Drive, generates a structured response with Google Gemini, and escalates low-confidence answers by creating a support ticket via an HTTP request.
How it works
- Receives an employee question via a POST webhook request.
- Fetches a policy PDF from a Google Drive download URL and extracts its text content.
- Analyzes the document text to detect the likely policy domain and document structure, then builds a system prompt that instructs Google Gemini to answer using only the document.
- Sends the prompt and the user’s question to Google Gemini and parses the model’s JSON response, including a confidence score.
- If the confidence is 0.70 or higher, returns the policy-based answer (plus section, next action, and summary) as the webhook response.
- If the confidence is below 0.70 (or parsing fails), creates a support ticket by POSTing the question and the bot’s best attempt to an external helpdesk endpoint and returns an escalation notice to the webhook caller.
Setup
- Configure the webhook endpoint in your chatbot or frontend to POST a JSON body containing at least
message,userId,channel, andsessionId. - Replace the Google Drive download URL in the PDF fetch step with your own policy document link that is accessible to n8n.
- Add a Google Gemini (PaLM) API credential and select an appropriate Gemini model.
- Set the helpdesk/ticketing system URL in the support-ticket HTTP request step and adjust the JSON payload fields to match your ticket API.
Additional info
Additional Possibilities
The same architecture can be adapted for:
- Employee handbooks
- SOP documentation
- Compliance manuals
- Vendor management guides
- Corporate knowledge bases
- Internal operating procedures
- Quality assurance documentation
and many other document-driven support use cases.
Troubleshooting Guide
| Issue | Possible Cause | Solution |
|---|---|---|
| Webhook returns no response | Workflow inactive | Activate workflow and use production URL |
| PDF extraction fails | Invalid PDF URL or permissions issue | Verify PDF is publicly downloadable |
| Gemini node errors | Missing or invalid credentials | Reconnect Gemini API credentials |
| JSON parsing error | AI returned invalid JSON | Review prompt configuration and Gemini output |
| Constant escalations | Confidence threshold too high | Lower threshold from 0.70 to a suitable value |
| Empty answers returned | Policy content missing required information | Verify PDF contains the requested information |
| Ticket creation fails | Incorrect helpdesk API URL | Verify endpoint and authentication settings |
| Wrong policy domain detected | Domain keywords do not match document content | Update keyword patterns in analysis node |
| Slow responses | Large PDF size | Reduce document size or implement retrieval methods |
| Employee receives escalation unexpectedly | AI confidence below threshold | Review policy coverage and prompt instructions |
Need Help?
Whether you need help with:
- n8n workflow setup
- Gemini AI integration
- Helpdesk integrations
- Slack or Teams deployment
- Policy document automation
- Multi-document knowledge bases
- Vector database integrations
- Enterprise AI assistants
- Hiring n8n developers
the team at WeblineIndia can help.
We can assist with workflow installation, customization, advanced integrations, performance optimization and development of similar AI-powered automation solutions tailored to your business processes.
If you need a custom version of this workflow or additional enterprise features, reach out to WeblineIndia for consultation and implementation support.
🔗 Nodes Used
Google Sheets, HTTP Request, Webhook, Extract from File, Google Gemini
📥 Import
Download workflow.json and import into n8n:
Workflow menu → Import from File