Project led by Víctor Muñoz, now an Akaen partner.
A leading food and beverage corporation in Spain managed thousands of points of sale and a critical volume of B2B orders. The bottleneck was not in demand, but in order entry: heterogeneous PDF documents, manual validation and delays before orders were entered into SAP.
Executive summary
About the customer
A multinational food company with logistics supplying thousands of outlets in Spain. Its priority was to accelerate the order cycle while preserving SAP data integrity.
The Challenge: heterogeneous orders and critical validation
Orders arrived in unstructured formats such as PDFs and emails, requiring manual reading to identify SKUs, quantities and delivery locations. Each order also had to be validated against SAP customer and product master data, creating a bottleneck during high-demand periods.
The solution: event-driven architecture with AI
We designed an Event-Driven architecture on Azure to process each order independently and at scale. The workflow combines cloud orchestration, cognitive extraction with GPT-4.1 and SAP validation before automatic entry into the ERP.
Automated Processes (3 key flows)
1. Intelligent capture and intake
The Problem: attachments in multiple formats requiring manual review.
Transformation: orchestration through Logic Apps, Blob storage and Service Bus queuing to absorb demand peaks without reducing performance.
2. Cognitive extraction of orders
The Problem: PDFs without fixed structure, difficult to map to SAP.
Transformation: Azure Function + Container Apps that convert PDFs into JSON with header, lines and units using GPT-4.1.
3. Validation and insertion in SAP
The Problem: operational risk if order data does not match customer or product master data.
Transformation: real-time API validation and a signed URL for the original PDF, providing an audit trail.
Operational comparison
- • Manual PDF reading.
- • Validation one by one in SAP.
- • ~15 min per order.
- • Queues during demand peaks.
- • Structured automatic extraction.
- • SAP validation in real time.
- • ~30 s per order.
- • Event-driven scaling.
The Impact
The shift from a reactive to a proactive model freed up critical administrative capacity and stabilised the supply chain during high demand. The order cycle became much shorter without compromising controls or traceability.
- Cycle time: from 15 minutes to 30 seconds per order.
- Operational capacity: absorbing peaks without expanding the team.
- Data quality: cross-validation with SAP before insertion.
Operational results
- Stronger SLA performance: more reliable response times for B2B customers.
- Total traceability: direct link to the original document from SAP.
- Actual scalability: architecture prepared for seasonal growth.
Conclusion
“Automation not only accelerates the order: it turns logistics into an autonomous and auditable system.”
Want to speed up your supply chain? .