← Cases
South KoreaTechnology and telecom2026 – ongoing

SK Telecom

In-call assist and post-call processing for customer service with Claude

SK Telecom's Claude call-center tools raise agent-rated answer quality by 34%

WorkingVendor's case studyVendor ↗

Problem

Call-center agents had to search documents while talking to customers, then write up each call afterward.

Approach

SK Telecom runs Claude on Amazon Bedrock, combined with its own retrieval model. During calls it supplies answers to agents, and afterward it summarizes, classifies topics, drafts to-do lists and rates sentiment, with human review.

Results

34%increase in human agents' quality ratings of in-call assistance answersVendor's case study[1]
68%fewer low-quality responses from the telecom fine-tuned model than the base modelVendor's case study[1]

Takeaway

Human agents reviewed the AI's live call assistance and post-call work. Anthropic reports post-call quality at about 89% of human agents, with quality ratings supplied by the agents using the tool.

Categories

Business function
Customer service
Innovation type
Process innovation
AI technology
Generative AI
Evidence
Vendor's case study

Sources

  1. SK Telecom customer story
    Anthropic2026Vendor publication

Source summaries

[1]Anthropic2026

Anthropic describes SK Telecom's use of Claude on Amazon Bedrock for live call assistance and post-call work. It reports quality gains rated by human agents.

Full details are in the original sources.

Added 25 Oct 2026 · Updated 25 Oct 2026
Open case page ↗