Fintech · Insurtech · 50–200 · North America #b009_04

Context

We build underwriting workflow software for property and casualty carriers, with language-model features embedded in how underwriters triage and price risk.

Industry

Fintech · Insurtech

Size

50–200 employees

Region

North America

Tools budget

$1K–5K per month

Needs

LLM evaluation & testing. We ship model-driven features to carriers and need systematic evaluation so prompt changes do not silently degrade underwriting answers.

Today: an AI systems group building internal prompt iteration and evaluation harnesses itself · Urgency: now

Guardrails for AI agents and LLM apps. Agents act on sensitive carrier data, and we need guardrails and audit trails customers will accept.

Today: customer-by-customer data isolation and manual review of what agents may do · Urgency: later

Who decides

cto · security review

Expects from suppliers

Evaluation that runs inside our delivery pipeline; Clear data isolation per enterprise customer; Tracing for multi-step agent runs

Not a fit

Consumer chatbot builders; Tools that require sending customer data to unknown regions

Stack

LLM APIs in production, CI/CD pipeline in place, cloud: AWS-class hyperscaler

Buying triggers

AI adoption push, new market or product launch