What challenges did Empower face before AI implementation?

Manual QA reviewed only a small subset of roughly 10 million annual calls. Scoring was inconsistent across analysts and feedback took days to weeks. The company needed to scale QA coverage without adding significant manual effort.

Source

ProductionEvidence: Medium50/100

Empower scaled contact center quality assurance with Amazon Connect Agent Evaluation API and Amazon Bedrock (20x QA coverage)

Use case typeCustomer service automationUpdated Jun 13, 2026

Empower, a financial services company serving over 18 million Americans, transformed manual contact center quality assurance using Amazon Connect Contact Lens and Amazon Bedrock with Claude 3.5 Sonnet. The solution processes thousands of pre-redacted call transcriptions per day, evaluates calls against the GEDAC framework, and writes standardized results back into Amazon Connect via the Agent Evaluation API.

Organization: Empower
Industry: Finance
Location: United States
Published: August 2025

Reported outcomes

QA call coverage increase: 20×

Other quantified impact

Production timeline: 7 monthsDaily transcription volume processed: 5,000 transcriptions/day

Planned next steps

The source says the organization aims to achieve: Freed staff for higher-value coaching.

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

QA call coverage increase: 20 x increase

AWS BlogAug 4, 2025Blog postExplicit claimMedium evidence strength

With this solution Empower saw a remarkable 20x increase in QA call coverage

Normalized claim

Production timeline: 7 months decrease

AWS BlogAug 4, 2025Blog postExplicit claimMedium evidence strength

delivered a production-ready generative AI solution from experiment to production in just 7 months

Normalized claim

Daily transcription volume processed: 5,000 transcriptions/day increase

AWS BlogAug 4, 2025Blog postExplicit claimMedium evidence strength

The solution processes 5,000 pre-redacted transcriptions per day in batches

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Empower
Provider: AWS
Maturity: Production
Linked source: AWS Blog

Empower and Accenture iterated prompts and deployed the system from experiment to production in about 7 months

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1Contact Center AI
2Quality Assurance Automation
3Workflow Orchestration

Built a production generative AI QA pipeline with Amazon Connect Contact Lens, Amazon Bedrock, AWS Step Functions, AWS Lambda, Amazon SQS, Amazon S3, and Amazon EventBridge.
Used Claude 3.5 Sonnet to evaluate calls against all GEDAC categories and delivered results into the existing Amazon Connect Quality Management UI.
Empower and Accenture iterated prompts and deployed the system from experiment to production in about 7 months.

Technologies

Amazon Connect Contact Lens Amazon Bedrock Claude 4.5 Sonnet AWS Step Functions AWS Lambda Amazon SQS Amazon S3 Amazon EventBridge Amazon Connect Agent Evaluation API

Architecture

Empower and Accenture built a production generative AI quality assurance pipeline around Amazon Connect Contact Lens transcriptions. New PII-redacted call transcripts are stored in Amazon S3, detected by Amazon EventBridge, queued in Amazon SQS, processed by AWS Step Functions and AWS Lambda, evaluated by Amazon Bedrock using Claude 3.5 Sonnet against the GEDAC framework, and then written back into Amazon Connect through the Agent Evaluation API for manager review in the existing Quality Management interface.

Implementation partners1

Accenture

Sources & evidence1

Evidence: Medium50/100Evidence strength