TruvioSense - Coding & Approvals
Coding & Approvals is an AI-powered extension for Microsoft Dynamics 365 Business Central that enhances ExFlow Accounts Payable by automatically predicting invoice coding based on historical invoice processing patterns.
This guide covers how to install, configure, and use TruvioSense in your daily invoice processing workflow.
What TruvioSense predicts
The system uses machine learning to suggest values for new invoices based on your historical coding patterns:
- G/L Accounts — account codes per invoice line
- Dimensions — Department, Cost Center, Project, and other configured dimensions
- Header fields — vendor, invoice number, currency, net amount, VAT amount
- Line-item details — quantities, unit costs, and descriptions
Key benefits
| Area | Benefit |
|---|---|
| Speed | Reduces invoice coding time by up to 80% |
| Accuracy | Improves coding consistency across your team |
| Risk | Reduces human error in account assignments |
| Throughput | Speeds up invoice approval and posting |
| Learning | Predictions improve automatically over time |
How TruvioSense works
The diagram below shows the end-to-end flow from setup to daily use.
One-time setup (administrator):
- Install TruvioSense - Coding & Approvals from AppSource
- Run data upgrade, onboard, and schedule the background job
- Enable companies and selected vendors for machine learning (referred to as ML throughout this guide)
Automatic background processing:
- Historical posted invoices and document images are exported to the TruvioSense cloud service via the Job Queue
- ML models analyze coding patterns and build vendor-specific prediction models
- Newly posted invoices are continuously exported to improve future predictions
Daily workflow (AP user):
- Open an invoice in ExFlow Import Journal and select View Predictions
- Review the coding suggestions — including header fields, line coding, dimensions, and confidence scores
- Apply predictions to the journal with one click
- Adjust any values manually if needed, then continue with your normal approval workflow
Your manual corrections feed back into the ML models automatically — no retraining action is needed.