Natixis CIB in New York improved invoice processing accuracy by reimagining its entire invoicing workflow with Microsoft Power Platform.

Previously, invoices arrived by email and were processed manually. A dedicated employee filtered invoice emails, keyed each invoice into the purchasing system, and submitted them for payment. This approach created:

• High workload and delays
• Limited controls and reporting
• Increased risk of human error

The team first tried building its own AI Builder custom component, but the trained model only reached about 60–70% accuracy. They then switched to the prebuilt AI Builder invoice processing model in Power Automate, which delivered a stable success rate of about 93.3% and helped the team reach around 90% accuracy in operational invoice processing.

Key elements of the new setup include:

• Power Automate to automatically extract invoice data from emails
• AI Builder prebuilt model for invoice recognition and data extraction
• Microsoft Dataverse to store structured invoice data
• Power Apps canvas app to monitor AI flow activity and adjust invoices when needed
• Power BI to visualize and analyze invoice data

This combination of tools significantly increased accuracy, reduced manual data entry, and gave the business more reliable invoice data to work with.

Automating invoice processing with Microsoft Power Platform helped Natixis CIB NY reshape its operations and reduce risk across the invoicing lifecycle.

Key business impacts include:

• Higher accuracy and fewer errors
  Using the AI Builder prebuilt invoice model, the team achieved about 90% accuracy in invoice processing, with a reported 93.3% success rate from the model itself. This reduced the risk of errors from manual data entry.
• Significant workload reduction
  Natixis CIB NY processes roughly 3,000–5,000 invoices per year. Automation reduced the workload of one full-time employee by more than half, freeing staff to focus on higher-value tasks instead of manual reconciliation.
• Faster, more efficient reconciliation and payments
  With automated extraction and centralized data in Dataverse, invoices move through the process more quickly, supporting timelier reconciliation and payments.
• Better visibility and control
  Power BI dashboards give senior management visibility into how many invoices are being processed and how they align with purchasing records and budgets. This helps identify gaps and supports more informed budget decisions.

Overall, the organization moved from a manual, opaque process to a more controlled, data-driven invoicing environment.

To manage and scale Power Platform effectively, Natixis CIB NY worked closely with the Tech Expertise and Solutions department at Groupe BPCE and adopted a structured approach to governance.

Groupe BPCE created a Power Platform center of excellence (CoE) with the goal of supporting the entire group. This CoE focuses on:

• Defining governance for how Power Platform is used across the organization
• Ensuring compliance with IT and security rules
• Providing training and support for developers and administrators
• Building a community user group to share best practices and explore new automation opportunities

For Natixis CIB NY, this meant they could:

• Confidently deploy Power Automate, Power Apps, AI Builder, Dataverse, and Power BI within corporate standards
• Continuously monitor AI flow performance and accuracy via a dedicated Power Apps canvas app
• Iterate on their invoice solution and look for additional operational tasks to automate

This governance model helped the organization not only improve invoice processing but also set a foundation to extend low-code and AI-driven automation to other business processes.