JapanFinance and insurance2025 – ongoing

Tokio Marine & Nichido Anshin Life Insurance

Generative AI classification of customer feedback

Tokio Marine & Nichido Anshin Life cut feedback processing time by about 75% with generative AI

WorkingVendor case studyVendor ↗

Problem

The insurer received up to 18,000 customer feedback entries a year. Staff read and categorized each one by hand, and their judgments on categories varied.

Approach

The insurer used IBM watsonx.ai with natural language processing and retrieval-augmented generation to sort feedback into about 40 industry-wide codes. A second phase added keyword extraction, hashtags and cluster analysis.

Results

about 75%reduction in time to read and categorize each feedback entry, from 7 to 8 minutes to 2 to 3 minutes, per IBMVendor case study[1]

about 500 hourswork hours saved per year, roughly 20 business days, per IBMVendor case study[1]

Takeaway

A fixed code list let the model apply the same categories every time, which removed differences between staff.

Sources

Promoting DX to utilize customer feedback in business operations through generative AI
IBM9 Dec 2025Vendor publication

Source summaries

[1]IBM9 Dec 2025

IBM reports how Tokio Marine & Nichido Anshin Life Insurance used watsonx.ai to classify customer feedback. Handling time per entry fell from 7 to 8 minutes to 2 to 3 minutes, and about 500 work hours a year were saved.

Full details are in the original sources.

Added 4 Nov 2026 · Updated 4 Nov 2026