What challenges did Insurancemarket face before AI implementation?

Manual review covered only a small fraction of customer calls because each call took 20-25 minutes to review. Limited sampling left coverage confusion, complaints, and emerging service issues unexamined and slowed response to recurring concerns.

What was the impact of the AI solution?

The company now analyzes virtually 100% of calls. Call analysis time dropped from 20-25 minutes to around 2 minutes. The system processes 400-500 calls in parallel per hour. Two employees were redeployed from manual call review. Coaching and training effectiveness is estimated to improve by 10-15%.

Source

Insurancemarket and Chaos Gears: AI-powered call analytics for near-real-time insurance support

Use case typeConversational analyticsUpdated 3 days ago

Insurancemarket, a Greek insurance aggregator, partnered with AWS partner Chaos Gears to build Voxflow, an automated call analytics system. The solution uses Amazon Bedrock, Amazon S3, AWS Lambda, and Amazon EventBridge to transcribe and analyze customer calls as they arrive. It provides near-real-time quality evaluation, sentiment analysis, structured reports, and follow-up workflows for supervisors and agents.

Organization: Insurancemarket
Industry: Insurance
Location: Greece
Published: July 2026

Reported outcomes

100%

call coverageOther quantified impact

2 minutescall analysis time450 calls/hourcalls processed in parallel per hour2 employeesemployee redeployment10%coaching and training effectiveness15%coaching and training effectiveness

Strategic outcomes

Customer experience & trustNear-real-time issue detection and follow-upBetter decisions & insightComplete view of call sentiment and agent performanceEmployee experienceFreed staff for higher-value work

Primary read

Use case focus

Showing 2 of 2

1Conversational analytics
2Workflow automation

Manual review covered only a small fraction of customer calls because each call took 20-25 minutes to review.
Limited sampling left coverage confusion, complaints, and emerging service issues unexamined and slowed response to recurring concerns.

Insurancemarket built a fully serverless call analytics workflow on AWS with Chaos Gears.
The system uses Amazon Bedrock with Anthropic Claude Sonnet for transcription analysis and quality evaluation, stores recordings in Amazon S3, processes each recording with AWS Lambda, and schedules the workflow with Amazon EventBridge.
Voxflow separates speakers, identifies topics, evaluates empathy, tone, and success, and sends structured reports to supervisors.

Technologies

Amazon Bedrock Amazon S3 AWS Lambda Amazon EventBridge

The company now analyzes virtually 100% of calls.
Call analysis time dropped from 20-25 minutes to around 2 minutes.
The system processes 400-500 calls in parallel per hour.
Two employees were redeployed from manual call review.
Coaching and training effectiveness is estimated to improve by 10-15%.

Architecture

Voxflow is a fully serverless call analytics pipeline that uses Amazon Bedrock with Claude Sonnet for call transcription analysis and quality evaluation, Amazon S3 for recording storage, AWS Lambda for processing each recording on arrival, and Amazon EventBridge for event-driven workflow scheduling.

Implementation partners1

Chaos Gears

Sources & evidence1

Groundedness: 5/5Type: Customer StoryPublished: Jul 15, 2026Publisher: AWSEvidence: PrimaryConfidence: High

Primary source

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Customer evidence