This Insurance Firm Gave 400 Staff AI Helpers for $14 Each

A German insurance broker called MRH Trowe has given about 400 employees their own AI assistants — and lets them build new ones themselves, no programmers needed. The rollout reached full company scale within the first month, at roughly $14 per person. The point wasn't flashy chatbots. It was letting staff ask questions, pull answers out of internal documents, and run multi-step tasks while everything stayed inside one controlled company system.

The problem it solves: employees were already experimenting with AI on their own. For a business handling sensitive insurance data in a heavily regulated industry, that's a nightmare — client details could leak into random public chatbots, and nobody would know what was asked or answered. The new setup combines three tools: one for building AI helpers, Amazon's service for running them safely and separating each session, and an open-source chat window workers already find familiar.

The catch: this is a vendor's own case study, so the numbers come from the companies selling the technology, not an independent audit. Also, "self-service" still means someone must set up rules, spending caps, and permissions — those budget limits exist precisely because AI bills can run away. And in a regulated industry, an AI answer is a draft, not a decision; a human still signs off on anything touching a client.

Why you should care: this is the template most workplaces will copy. Instead of banning AI, companies are buying it in bulk, cheaply, and locking it down. That means your next job likely includes AI tools you're expected to use — and a per-seat bill your employer watches closely.