Kapil Chauhan

Professional indemnity insurance

Inperio: AI underwriting in production

An AI underwriting platform for insurance submissions, regulatory evidence, claims context, and human-reviewed decisions.

Open Inperio

AMIE shared memory settings for firm context, with account details and policy values anonymized
AMIE Memory lets administrators maintain firm context that the assistant references in conversations. Account details and configured policy values have been anonymized in this image.

Context

Underwriters reconcile proposal documents, claims history, rate cards, prior decisions, and regulatory evidence. Manual review is slow and needs traceable evidence.

System and approach

AMIE ingests and routes email and upload submissions, extracts structured facts, cross-checks relevant regulatory and company sources, applies governed and versioned underwriting rules, and presents cited analysis, pricing context, discrepancies, and draft outputs for human review.

  • Ingest and route proposal and claim submissions.
  • Cross-check proposal facts with SRA, FCA, and Companies House, plus claims and prior decision context.
  • Retain source citations and surface discrepancies and pricing rationale.
  • Keep decisions, workflow changes, and outbound drafts under role-based access and explicit human approval.

Delivered scope

Production web application with underwriting inbox and board, analysis tickets, regulated directories, pricing support, document export, Salesforce workflow, and audit controls.

Outcome

Production outcome: £20,000 per month in operating savings. Underwriters remain responsible for final judgments, and exceptions stay reviewable.

Technology

  • Python
  • Flask
  • React
  • PostgreSQL
  • Celery
  • Redis