ClearShield Insurance — ClearShield Insurance is one of the UK’s top-20 general insurers, writing over £900M in annual premiums across home, motor and commercial lines, and processing approximately 180,000 claims per year.
Faster claims processing
Monthly operational savings
Straight-through processing rate
Net Promoter Score improvement
The Chalange
ClearShield’s claims operation relied on a patchwork of legacy systems, spreadsheet-based tracking, and manual adjudicator workflows. Average claim cycle time was 14 days for motor and 22 days for home claims. Customer satisfaction scores were declining, fraud detection was inconsistent, and adjudicators were overwhelmed with low-complexity claims that could theoretically be auto-settled. Recruitment pressure was intensifying the backlog.
What did
Automate Stacks do
Automate Stacks implemented a three-tier automation architecture starting with RPA for data-heavy manual tasks, layering in AI-powered document processing, and culminating in an Agentic AI claims triage engine:
RPA Deployment: 18 attended and unattended bots were deployed to handle FNOL data capture, policy lookup, reserve setting and payment initiation across ClearShield’s three legacy claims platforms.
AI Document Processing: An IDP model was trained on 200,000 historical claims documents to extract key fields from loss adjuster reports, repair estimates, medical certificates and police reports with 97.4% field-level accuracy.
Agentic Triage Engine: An autonomous agent evaluated incoming claims against fraud indicators, policy terms and claim history, routing 91% of standard claims to straight-through processing with zero human touchpoints.
The Results
- Claims cycle time reduced from an average of 14 days to under 3 days for motor claims
- 91% straight-through processing rate achieved for standard claims
- £750,000 per month in operational cost savings within 6 months of deployment
- NPS improved by 28 points due to faster settlements and proactive customer communication
- Fraud detection accuracy improved by 44% through AI pattern analysis at triage
