In Property & Casualty insurance, claims speed directly affects customer retention and loss ratios — delays compound into complaints, litigation risk, and churn. Tokio Marine Insurance UAE's claims team was manually processing up to 600 car insurance claims per day, each requiring extraction of dozens of data points from documents including police reports, Emirates IDs, driver's licences, and vehicle registrations. Every submission demanded repetitive, error-prone manual data entry under significant time pressure. The 2024 UAE floods pushed volumes to record highs, exposing the fragility of a process that had no capacity headroom and was generating costly backlogs and delayed customer communication.
Tokio Marine Insurance UAE partnered with Kodak Alaris and deployed its Info Input platform, integrated with Microsoft Document AI and powered by Natural Language Processing (NLP). The engagement began with structured discovery sessions to map document types and extraction requirements, followed by a proof of concept to validate accuracy before full deployment. The production system automatically ingests claims submissions arriving via email, applies NLP-based extraction to parse and validate key data fields across multiple document formats, and routes exceptions for human review only when confidence thresholds are not met. This replaced end-to-end manual data entry with automated ingestion and validation, integrating directly into the existing claims workflow without requiring staff to change how submissions are received.
Claims processing time dropped by 90%, enabling customer communication often within hours rather than days. The system now handles 3x the previous daily claims volume without additional headcount, effectively tripling throughput capacity. Exception handling decreased by 60%, reflecting the accuracy gains from automated validation over manual keying. Additional qualitative outcomes include:
Have a similar implementation?
Share your customer's AI results and link it to your vendor profile.
Submit a case study →