Classify purchase requisitions with Google Gemini, Gmail, Slack and Sheets

Quick overview

This workflow receives purchase requisitions via a webhook, validates and normalizes the payload, uses Google Gemini to classify the requisition into a procurement category with confidence and risk signals, then routes for auto-approval or manual review, logs results to Google Sheets, and notifies stakeholders via Gmail and Slack.

How it works

1. Receives a new purchase requisition as a POST request to a webhook endpoint.
2. Validates required fields (PR ID, requester, description, amount) and normalizes the requisition data into a consistent structure.
3. Sends the requisition details and allowed category list to Google Gemini and requests a strict JSON classification result (category, confidence, risk, correction, reasoning).
4. Parses the AI JSON response and derives an approval decision based on confidence thresholds and misclassification risk.
5. If eligible, marks the requisition as auto-approved and ready for ERP routing; otherwise, marks it as pending manual review and emails a review ticket via Gmail.
6. Combines the original PR data with the classification outcome, appends an audit record to Google Sheets, and sends status notifications via Gmail and a Slack channel.

Setup

1. Configure the webhook URL (POST /pr-submission) in your PR submission source and ensure it sends pr_id, requester, description, and amount in the request body.
2. Add Google Gemini (PaLM/Gemini) API credentials for the categorization step.
3. Add Gmail OAuth2 credentials and set the recipient addresses for the manual review email and the final processing notification.
4. Add Slack OAuth2 credentials and select the target channel for procurement alerts.
5. Add Google Sheets credentials, create/select the target spreadsheet and PR_Records sheet, and ensure the columns match the fields being appended (for example pr_id, final_category, confidence_score, and approval_status).

Additional info

How To Customize Nodes

Customize Procurement Categories

Modify the allowed categories inside:

• AI: Categorize Purchase Requisition

Current categories:

• IT Equipment
• Office Supplies
• Software
• Facilities
• Marketing
• HR & Training
• Professional Services
• Logistics
• Uncategorized

Adjust Confidence Thresholds

Modify logic inside:

• Calculate Classification Confidence

Current approval logic considers:

• Confidence score
• Misclassification risk
• AI approval recommendation

Typical behavior:

• High confidence + low risk → Auto Approved
• Medium confidence or elevated risk → Manual Review
• Low confidence or category mismatch → Manual Review

Modify Approval Logic

Update the node:

• Check Auto Approval Eligibility

You can add additional business rules such as:

• Department-specific approvals
• Amount-based approval routing
• Vendor-based escalation

Customize Notification Templates

Edit message templates inside:

• Gmail nodes
• Slack node

You can include additional PR fields or approval details.

Extend Google Sheets Schema

Add more tracking columns in:

• Store Procurement Record

Examples:

• Vendor Name
• Cost Center
• Priority Level
• Procurement Owner

Add-ons

You can further enhance this workflow with:

• ERP integration (SAP, Oracle, NetSuite, Dynamics)
• Approval workflows using Microsoft Teams or Slack actions
• Database storage using PostgreSQL or MySQL
• Vendor risk scoring
• OCR extraction from invoices or attachments
• Multi-level procurement approvals
• AI-powered fraud detection
• Procurement analytics dashboards
• Automated PO generation
• Integration with ticketing systems like Jira or ServiceNow

Use Case Examples

1. IT Procurement Automation

Automatically classify laptop, monitor, and hardware requests into the IT Equipment category.

2. Office Supply Management

Categorize recurring office supply purchases without manual procurement review.

3. Procurement Compliance Monitoring

Detect potential category mismatches and route suspicious PRs for manual review.

4. Enterprise Procurement Audit Tracking

Store all procurement decisions, confidence scores, and AI reasoning for compliance and auditing purposes.

5. High-Volume PR Processing

Reduce procurement processing time by auto-approving high-confidence purchase requests.

Additional use cases can be built depending on your procurement policies, ERP ecosystem, and approval structure.

Troubleshooting Guide

Issue	Possible Cause	Solution
Webhook not receiving requests	Incorrect webhook URL	Verify webhook endpoint configuration
AI categorization fails	Google Gemini credentials missing or invalid	Reconnect Google Gemini credentials
PR rejected during validation	Required fields missing	Ensure pr_id, requester, description, and amount are included
Gmail notifications not sent	Gmail OAuth2 not configured	Reconnect Gmail credentials
Slack alerts not delivered	Invalid Slack channel ID	Update the Slack channel configuration
Google Sheets insertion fails	Incorrect spreadsheet or permissions	Verify spreadsheet ID and sharing permissions
AI response parsing error	Invalid JSON returned by AI	Review AI prompt formatting
Requests always routed for review	Confidence thresholds too strict	Adjust confidence scoring logic
Duplicate PR records	Same PR submitted multiple times	Add duplicate detection logic
Workflow execution stops unexpectedly	Node configuration issue	Check execution logs in n8n

Need Help?

If you need assistance with:

• Setting up this workflow
• Customizing procurement logic
• ERP integrations
• Multi-level approval systems
• Advanced AI procurement automation
• Dashboard creation and analytics
• Workflow scaling and optimization

the n8n workflow development team at WeblineIndia can help you build, customize and deploy enterprise-grade n8n automation workflows tailored to your business requirements.

You can also extend this workflow with additional integrations, approval chains, procurement intelligence and enterprise compliance features based on your operational needs.