Case study · Insurance

Agents that work the inbox.

A broker team that receives hundreds of questions by email and WhatsApp every week.

Illustration: Insurance

The challenge

Answers depended on looking up policies in several systems. Response times grew with volume and the team spent the day copying information.

Our approach

  1. Read the channels

    An AI assistant reads incoming email and WhatsApp messages and identifies the client and policy.

  2. Query the systems

    It checks the policy systems and prepares a draft answer with the right data.

  3. Keep a human in the loop

    Brokers review, adjust and approve every answer before it is sent.

What changed

  • Draft answers ready when the broker opens the message.
  • Policy data pulled automatically instead of copied.
  • Every reply still reviewed by a person.

Stack

  • AI agents
  • MCP
  • TypeScript
  • Email
  • WhatsApp API
  • Human review