AI-Powered CX

Unpacking my ‘go-to’ architecture for delivering web-scale customer interaction models.

Certificationn Badge Certificationn Badge Certificationn Badge Certificationn Badge Certificationn Badge
Certificationn Badge Certificationn Badge <
Certificationn Badge Certificationn Badge Certificationn Badge
Note: This page is a work in progress; please forgive incomplete descriptions.

This architecture reflects my current technical acumen and business philosophies:

  • Customer is most important person in the room
  • Do more with less
  • Secure by design
  • Learn & apply
  • Follow the data
  • Simplicity accelerates

I use it within my own projects — and as a baseline for related consulting work.

It’s an ideal — secure, performant and modular — platform for delivering:

  • Personalization
  • High-scale engagement
  • Conversational UI

Inspired — and perpetually evolving — from AWS reference architectures and experience design best-practices:

  • Anticipation
  • Consistency
  • Omnipresence

And continuously validated against the AWS Well-Architected and NIST Cybersecurity frameworks.

Push-Button Deployment

Maintained in a library of CloudFormation templates and deployable on-demand — as individual modules or end-to-end.

Design Considerations

  • Why not TerraForm? CloudFormation covers everything required in a universal format (YAML/JSON), so there is no compelling reason to take on the added costs and complexity of a 3rd-party tool
  • Why AWS? In a word…ecosystem. The breadth and quality of services, coupled with their integration, provide the most compelling business case. And after working with nearly all of AWS’s services over the past 10+ years — and applying their published best practices and reference architectures — AWS has earned my trust…and my business. I’ve also used GCP and Azure quite a bit, but have always ended up back on AWS due to performance or architectural reasons.

The Bigger Picture

This architecture also fills a key role in a more comprehensive cloud strategy:

So let’s dive in…

Core Components

Laying the groundwork for world class AI-powered customer interfaces.

In our modern information-rich economy, attention is the core commodity. And consumer attention is won with context. While context is created with personalized experiences.


  • Sentiment analysis
  • Image/video recognition
  • Form processing and list management

Expansion Modules

With this solid foundation in place, we can easily enhance functionality using a number of different add-on modules:

  • Agent
  • Contact Center
  • Inference

Let’s dive a little deeper into those modules…

Agent Service

This add-on module provides a chatbot that can be integrated with application.

  • Lex — Natural language processing
  • Alexa — (optional) Natural language processing within Alexa ecosystem
  • Lambda — Core logic for bot
  • DynamoDB — Cross-session persistence and conversation logging
  • CloudWatch — Collect and store logs from Lambda

Direct integration with 3rd-party messaging platforms, e.g. Messenger, etc.

Contact Center Service

This add-on module enables a microservice for intelligent contact flows.

  • Connect — Central management of customer contact routing
  • Lex — Natural language processing
  • Lambda — Serverless computation
  • CloudWatch — Collect and store logs from Lambda

Voice-based customer service at web scale.

Inference Service

Add-on microservice providing custom decision automation.

Custom machine learning models, for example:

  • Product Recommendation
  • Content Recommendation

Translation Service

Add-on microservice providing language translation automation.

Application and website content.