AI propels underwriting and risk profiling: GlobalData poll points to biggest impact as Allianz's BRIAN saves 135 days
Insurers see near-term AI gains in underwriting and risk profiling, boosting decision speed and pricing accuracy. Allianz's BRIAN freed 135 days by answering requests at scale.
Underwriting and risk profiling: the near-term AI opportunity
A Q3 2025 GlobalData poll shows 45.8% of respondents expect underwriting and risk profiling to see the greatest upside from AI over the next five years. The signal is clear: insurers see AI improving decision speed and pricing accuracy where it matters most.
This shift is already visible in practice. Allianz has deployed BRIAN, a generative AI assistant in commercial underwriting, after a pilot that handled nearly 3,000 questions from 190 users. Since going live in January 2025, the tool has freed an estimated 135 working days by handling information requests with accurate, consistent answers.
What this looks like on the desk
- Submission intake and triage: auto-extract key fields, flag missing data, route to the right team.
- Knowledge retrieval: instant answers from approved sources (wordings, endorsements, appetite, prior decisions) with citations.
- Risk data enrichment: pull third-party data (geospatial, cat models, telematics) and summarize for quick review.
- Pricing support: compare similar risks, highlight material differences, surface rate adequacy signals.
- Portfolio steering: identify accumulation hotspots and referral triggers before quoting.
- Documentation: first drafts of quotes, endorsements, and broker emails, ready for underwriter edit.
- Audit and consistency: log rationale, sources used, and decision steps for QA and compliance.
Why junior underwriters benefit
Faster answers shorten the learning curve and reduce variance in decisions. With routine queries handled, juniors spend more time on risk analysis, pricing refinement, and broker relationships-skills that move the needle on hit rate and loss ratio.
Metrics that prove value
- Quote turnaround time and underwriter hours saved per submission.
- Hit and bind ratios by segment; quote-to-bind cycle time.
- Loss pick variance versus actuals; rate adequacy and leakage.
- Quality indicators: accuracy of extracted fields, citation precision, and exception rates.
- Compliance: completeness of audit trails and reduction in manual rework.
How to pilot with discipline
- Pick one high-volume, rules-heavy workflow (e.g., submission triage or knowledge retrieval) and define a clear success metric.
- Use a controlled knowledge base (policy wordings, endorsements, appetite guides, underwriting rules) with source citations.
- Keep a human-in-the-loop for approvals; start read-only before binding decisions use AI outputs.
- Backtest against historical cases; compare decisions and outcomes to baseline.
- Establish guardrails: data privacy, PII handling, prompt logging, and model risk management.
- Plan for scale: API integration to core systems, role-based access, and ongoing monitoring.
- Upskill teams with short, role-specific training and playbooks for consistent use.
What insurers should do next
Focus on high-impact tasks where AI reduces manual effort and decision friction. Pilot quickly, measure productivity and accuracy, then scale what works. As more carriers adopt similar tools, speed and consistency in underwriting will become a clear competitive edge.
If you're building capability across underwriting and pricing teams, explore practical learning paths and tool stacks:
Source: GlobalData poll (Q3 2025). Case example: Allianz BRIAN deployment details as summarized above.
