ONE Insurance

As Europe's first fully digital insurance provider, ONE Insurance had staked its competitive position on end-to-end digital processing across motor, home (personal contents), and personal liability lines. Yet beneath that digital promise, application intake still depended on manual keying and validation by brokers and staff. In Property & Casualty, application volume fluctuates with market cycles and geographic expansion, making manual-dependent workflows a structural ceiling on growth. Errors introduced during data entry compounded downstream in underwriting, raising correction costs and slowing policy issuance. With a triple-digit growth trajectory and active plans to expand across European markets, the bottleneck threatened both service quality and the unit economics required to scale.

ONE Insurance selected the Hyperscience Platform to automate data capture, classification, and extraction across the full range of document types encountered in P&C intake — including handwritten forms, low-resolution scanned images, and electronically generated PDFs. Hyperscience's IDP layer applies machine learning and NLP to convert unstructured inputs into machine-readable data, routing clean records into downstream underwriting and policy management systems. The platform deploys on-premise, learning continuously from ONE Insurance's own document corpus behind its firewall — a deliberate choice to improve accuracy over time without exposing sensitive policyholder data to external systems. ONE Insurance specifically selected Hyperscience for its proven ability to handle handwriting and low-resolution image quality reliably, its quick time-to-value, and its track record in the insurance industry.

The Hyperscience deployment enabled ONE Insurance to reduce reliance on manual keying and validation across all three product lines, improving data quality entering underwriting workflows and accelerating application processing at scale. Broker and employee capacity previously absorbed by repetitive data entry was redirected toward customer service and revenue-generating activities.

Key outcomes:

• Reduced manual keying and validation across motor, home, and personal liability lines
• Improved data quality flowing into downstream underwriting and policy management systems
• Freed broker and staff resources to focus on higher-value customer and commercial activities
• Scalable processing capacity supporting the company's triple-digit growth into broader European markets

• Handwriting and low-resolution image recognition quality should be a primary evaluation criterion when selecting IDP vendors for P&C insurance — most incoming documents are far from clean.
• On-premise ML deployment that learns from proprietary document data consistently outperforms static rule-based OCR in accuracy over time.
• Automating document intake early in the application workflow decouples headcount growth from volume growth, which is essential for insurers pursuing geographic expansion.
• Vendor track record in insurance specifically matters — domain-tuned models reduce time-to-value compared to general-purpose automation platforms.
• Freeing broker capacity from data entry creates compounding returns: cost savings stack with improved customer responsiveness.