Corvus Insurance achieves sub-40% loss ratio with generative AI-powered underwriting automation

Corvus underwriters spent significant time on routine manual tasks including industry classification research, manual data entry from insurance applications received by email, and cross-referencing application answers against complex underwriting guidelines. These activities reduced the time available for high-value work that drives growth and book value for brokers and risk capital partners.

Corvus added three generative AI and NLP-driven capabilities to its Corvus Risk Navigator™ platform: (1) Automated Industry Verification using a large language model to replace manual industry classification research; (2) Automated Application Intake to ingest security control question answers from emailed applications, eliminating manual data entry; and (3) Instant Guideline Validation to automatically check application responses against underwriting guidelines in real time.

The enhancements further reduced underwriter workload, increasing quoting speed and efficiency while the company maintained an industry-leading loss ratio below 40%. By automating routine tasks, each underwriter can spend more time on activities that drive value for brokers and risk capital partners, supporting higher growth and greater book value.

• LLM-powered industry classification can replace a time-consuming manual research step that underwriters perform on every submission.
• Automating application intake from unstructured email content (including security control Q&A) is a high-ROI use case for NLP in specialty insurance workflows.
• Combining automation of routine tasks with strong underwriting fundamentals (reflected in a sub-40% loss ratio) demonstrates that AI augmentation and underwriting quality can reinforce each other.