Challenge
FM, one of the world’s largest commercia and industrial property insurers, evaluates thousands of broker submissions a year across more than 30 global offices. Each arrives by email as inconsistent, unstructured attachments: spreadsheets in dozens of formats, ACORD forms, and scanned PDFs, often in multiple languages. Underwriters had to triage, extract, and re-key every field into FM’s risk-scoring service before an account could even be evaluated. That administrative work slowed responses to brokers and meant strong risks could be missed or declined late.
Solution
Nimble Gravity designed and built FM’s AI Powered Submissions natively on Microsoft Azure AI Foundry inside FM’s own tenant. Six specialized agents handle intake, extraction of 300+ fields across50+ spreadsheet formats, enrichment, scoring through FM’s risk models, report generation, and broker communication. Underwriters review only the exceptions in a purpose-built human-in-the-loop console, and confidence gates route anything uncertain to a person. The platform passed FM’s EnterpriseArchitecture Review Board and went from contract to production in 24 weeks.
Results
100% of submissions scored on arrival
Every broker email is read, enriched, and scored as it lands, so underwriters start from a ranked, cited report instead of a raw inbox
65% faster on the most complex submissions
10-15% faster overall, with international, multi-language submissions now processed as fast as domestic accounts
From 20 to 130+ underwriters across 9 countries
Adoption grew from a January pilot to production underwriting teams worldwide
One platform, now extending to claims and impairment
The same agents, orchestration, and review console are the foundation for FM’s next AI workloads
