Six use cases in production, not pilots.
The mandate was the easy part; the second act is a field asking “what do I actually use it for?” Qumis hands you six answers your brokers already recognize — each with production evidence — running alongside the assistants you’ve already bought.
Architecture walkthrough plus a live working session on your documents.
Co-Lab — a loss run in, a client-ready claims dashboard out, every figure cited
Representative screen · specimen clientBreak down these loss runs for the renewal meeting — frequency vs. severity by policy year, the causes driving incurred, and anything the client should hear before we go to market.
Qumis · Claims specialist
- Read the loss runs — 5 carriers, 5 policy years0:18
- Normalized 64 claims to one schema0:24
- Computed frequency, severity, and development0:31
- Built the interactive claims dashboard0:52
The dashboard is saved to outputs/hartline_claims_dashboard.html — overview cards, the policy-year trend, a filterable claims register, and a deductible simulator.
1. Executive summary
Across five policy years Hartline has incurred $2,010,750 on 64 claimsSummary · r.4, and 2025–26 alone accounts for 41% of it.
- Paid & ALAE: $1,184,300 (58.9% of incurred)Paid · r.66
- Open reserves: $826,450 — two products claims hold 61%Open · r.3–4
- Average severity: $31,418, up from $23,556 in 2021–22
Ask Qumis anything…
*Use references provided to verify information.
Hartline Manufacturing
Claims analysis & risk insights · policy years 2021–2026
Adoption follows named work.
Turn the AI mandate into a business result.
Announced initiatives now need production wins. Six named use cases — each with a countable output and a recognizable owner — are an adoption map, not another pilot.
Solve “what do I use it for?”
Rollouts stall because people need to be told what to do in their own job’s terms. The six are exactly that: the internal enablement library you were going to have to write.
Coexistence, not a bake-off.
General assistants plus a specialist system — with the citation, abstention, and reliance discipline that client-facing insurance work requires of anything you approve.
Six named jobs, and the bridge to the assistants you already run.
Our largest customers run general assistants and Qumis side by side.
All six use cases
Named jobs with owners: service runs checking, claims runs analysis, placement runs towers, producers run proposals.
See the use case 02Why Qumis — the architecture
The specialist bench, a coverage attorney’s reading order, verified citations — what purpose-built means here.
See the use case 03Qumis MCP — beta
Qumis coverage analysis inside the assistants your firm already runs. In beta with design partners, the coexistence bridge.
See the use caseOur largest customers run enterprise deployments of general AI assistants and 100+ active Qumis users at the same time — coexistence in production, not in a slide.
Qumis production data, largest brokerage deployments
Underneath the six is Coverage Intelligence — the system of judgment Qumis is building for commercial insurance, where every reviewed decision makes the next one better.