United StatesFinance and insurance2024 – ongoing

Unum Group

Generative AI policy search for AskUnum employees

Unum's policy-search assistant returned results in 4 to 5 seconds, Microsoft reported

WorkingVendor case studyVendor ↗

Problem

AskUnum representatives manually searched digitized insurance policies while answering client inquiries. Finding relevant terms consumed time and created opportunities for mistakes.

Approach

Unum built a search application using Azure OpenAI Service and Azure AI Search, with support from Microsoft architects. The team refined retrieval and prompts during its pilot, then deployed the application for support employees.

Results

4 to 5 secondsaverage search response time, per MicrosoftVendor case study[1]

95%accuracy at launch, reported by Unum's IT delivery executiveCompany's own claim[1]

more than 90%AskUnum employees using the application, per MicrosoftVendor case study[1]

Takeaway

The reported speed concerned policy retrieval. It did not measure the time needed to resolve an entire customer inquiry.

Sources

Unum Group builds custom AI application to search 1.3 terabytes of data with 95% accuracy using Microsoft Azure generative AI solutions
Microsoft23 May 2024Vendor publication

Source summaries

[1]Microsoft23 May 2024

Microsoft described Unum's deployed policy-search application. It reported response times of 4 to 5 seconds and adoption above 90%, while quoting an executive's 95% launch-accuracy claim.

Full details are in the original sources.

Added 1 Nov 2026 · Updated 1 Nov 2026