Industry

Healthcare & AI/ML

Technologies Used

Amazon Connect
AWS Bedrock (Claude)
Amazon Transcribe
Amazon Polly
Amazon DynamoDB

"Chetu has been there with us. We've gone through years of not getting a good AI agent. Now we've got a good one. We've got all the supporting architecture. And, we've got the bot in test, and hopefully by the beginning of next month, it'll be all the way through our contact center. So far, the results have been great."- John A. Ross, CEO

Project Overview

Concierge Contact Center, a dental answering and receptionist services provider, partnered with us to modernize its Amazon Connect setup. Our experts at Chetu stabilized workflows, replaced third-party integrations with AWS-native services, and built an AI-driven automation pipeline featuring a beta voicebot and production roadmap. This transformation will allow Concierge Contact Center to turn its contact operations into a competitive advantage—delivering smarter, faster, and more personalized patient experiences.

The Client

Concierge Contact Center delivers 24/7 live agent support, live chat, scheduling, bilingual (English–Spanish) assistance, and HIPAA-compliant message capture for dental practices—ensuring reliable, secure patient communication. To strengthen coordination and visibility across its separate AWS environments for development, testing, and production, the leadership team aimed to modernize its infrastructure and accelerate adoption of an AI-driven operating model.

They set clear goals for their next phase of growth—stabilize operations, increase efficiency, and accelerate how quickly common requests are handled. Adopting AWS-native services and AI was central to that plan. This would enhance agent-assist and unlock real-time insights--without disrupting day-to-day operations.

Reflecting on the turnaround, John A. Ross shared, "I was skeptical after prior vendors, but Chetu followed the schedule to the day and even finished a project two weeks early; we now see them as a partner for maintenance and future development."

The Approach

Chetu was addressing high-priority JIRA tickets within milestone sprints and meeting deadlines always. The resolution substituted disjointed third-party tools with AWS-native parts and overlaid AI intelligence:

Call orchestration with Amazon Connect.

Amazon Real Time speech-to-text Transcribe.

AWS Bedrock (Claude) to intent routing and generate responses.

Amazon Polly to use high-quality text-to-speech.

Streaming, persistence, and search streaming, persistence and search with Amazon Kinesis, DynamoDB and OpenSearch.

The voicebot and agent-assist processes were created to benchmark cost-to-serve, CSAT, and containment-based on results that the adoption of AI agents on a large scale can result in up to a 50% cost-per-call decrease and CSAT improvement in top programs.

The Solution

We delivered a unified AWS-native foundation that replaced unreliable components and improved visibility and automation—providing a stable base to test and refine the AI voicebot and agent-assist features now in beta.

Centralized on an AWS-native contact center based on Amazon Connect with KMS, VPC, S3, DynamoDB, OpenSearch, QuickSight, CloudWatch, WAF, ELB, API Gateway, and SNS.

Developed an AI voicebot and agent-assist toolchain using Transcribe, Bedrock, and Polly, and automated escalation to human agents.

Implemented a data and analytics pipeline based on Kinesis streaming, Glue ETL, OpenSearch indexing of transcripts and performance insights, and Redshift and QuickSight reporting.

The Impact

The initiative produced measurable operational lift with a fully functioning beta AI voicebot on a scalable architecture, hitting every milestone on time. Looking ahead, industry research from McKinsey suggests AI adoption can achieve step-change efficiency--on the order of 40-50% fewer agents while handling 20-30% more volume as automation scales--expanding self-service for routine intents and sharpening agent-assist for complex cases.

Foundational analytics and integrations now fuel ongoing improvements, using tools like a bill payment calculator and informed CRM-versus-database decisions to enhance cost-to-serve, CSAT, and containment.

As John A. Ross put it, "We spent hundreds of thousands over three years trying to get an AI agent and couldn't get it done; Chetu went to the heart of the problem and solved it--we've got an AI bot in beta working well."

AI for Concierge Contact Center: Working Voicebot, Streamlined Flows, AWS-Native Stack