The context, the need 📍
Every year, Eurostar receives several thousand claims for reimbursement, accompanied by a variety of supporting documents (invoices, receipts, boarding passes, etc.). Until 2025, these documents were examined manually by experts, to ensure compliance and rigorous data entry.
This reliable but time-consuming method considerably lengthened processing times.
To increase efficiency, Eurostar needed an automated solution capable of :
- Automatically file different types of documents (invoices, boarding passes, receipts)
- Extract key information (dates, passenger names, amounts) with minimal human intervention
- Frictionless integration with its Salesforce-based case management system
The approach, solutions and key success factors 🛠
To meet these challenges, the ILLUIN Technology team has deployed ILLUIN Automateour configurable document AI platform, with the activation of the following functionalities :
Parameter-based classification engine
- An AI engine that can be parameterized without a complex training phase
- Quick set-up from less than 100 production documents
- Ultra-fast inference for high-volume environments
Tailor-made information extraction pipelines
- Combination of deep learning models and heuristic rules to automatically extract key fields: names, dates, amounts, reservation references, etc.
- Built-in compliance checks to detect inconsistencies and omissions
Salesforce integration
- Bi-directional API connectors automatically inject results (classification, extracted data) into reimbursement files
- Seamless integration into existing CRM ecosystems
Agile delivery & continuous improvement
- Project conducted in two-week sprints, in close collaboration with Eurostar teams
- Progressive roll-out of functionalities, adjusted to feedback from the field
- Continuous learning loop, fed by corrections validated by operators, guarantees constant performance improvement
The technical stack 🤖
Document AI solution :
Frameworks & tools :
Infrastructure :
The results, the benefits ✨
Since it went into production in Q1 2025, the partnership has delivered a significant return on investment, as well as major operational gains:
High precision and reliability
- 99% accuracy in document classification
- Over 90% reliability in data extraction
- Significant reduction in manual errors and re-entries
Major efficiency gains
- 80% reduction in processing time
- Agents now focus solely on validation, without manual input
Scalability and performance
- Capacity to process more than 5,000 documents per day without loss of quality or slowdowns
- Resilience in the face of peak activity
Enhanced customer experience
- Project Shortened processing times
- More consistent treatment, resulting in higher customer satisfaction
Enhanced fraud detection
- Structured, centralized data, facilitating rapid identification of potential anomalies or frauds
Conclusion
By automating document processing with ILLUIN Automate, Eurostar has considerably speeded up the management of its claims, making them more reliable.
The solution integrates seamlessly with existing systems, enhancing the customer experience and freeing up staff time.
This project is a concrete demonstration of how AI can modernize traditional processes and improve the customer experience, positioning Eurostar as a benchmark player in operational excellence and digital innovation.
Useful links 🔗
⚙️ Discover the ILLUIN Automate platform
💡 O ur custom-made projects in AI, GenAI & Data Science
