AI in Amazon Connect

One Contact Center, Three AI Engines.

How Bedrock, Lex, and SageMaker work together.

AI is transforming customer service, but figuring out how it actually fits into Amazon Connect can feel like drinking from a firehose. Amazon Lex, Amazon Bedrock, and Amazon SageMaker each do something different, and most deployments only need one or two of them working together, not all three from day one.

Request a Fit CheckSee the Comparison
3
AWS AI services that can power one Amazon Connect deployment
0
Model training required to start, with Bedrock and a Knowledge Base
5-20%
Typical reduction in Average Handle Time
Bedrock, Lex, and SageMaker connected within an Amazon Connect contact center

Standard Consulting Partner since 2017

The Big Picture

Four services, four different jobs

Today's contact centers get a serious AI upgrade with tools like these, instead of a static “Press 1 for Sales” menu.

Amazon Lex

The conversational interface, your bot's voice or chat, the layer that actually talks with a customer in natural language.

Amazon Bedrock

Access to powerful large language models like Claude and Titan, most often paired with a Knowledge Base for grounded, self-updating answers.

Amazon SageMaker

The build-your-own workshop for custom machine learning: predictions, fraud detection, sentiment scoring, anything an off-the-shelf model can't do.

Amazon Q

The newest generative AI assistant, plugging directly into Connect, often paired with the same Bedrock Knowledge Base underneath.


When To Use Which

Bedrock and a Knowledge Base, or a custom SageMaker model?

Most organizations only need one of these to start. Here's the honest comparison, not a sales pitch for the more expensive option.

Start Here Bedrock + Knowledge BaseCustom Model (SageMaker)
SetupPlug-and-play, no training neededFull ML pipeline setup
UpdatesAuto-syncs with new dataRequires retraining
CostPay-per-usePay for compute time and hosting
Best ForFAQs, self-service bots, knowledge lookupPredictions, analytics, custom use cases
MaintenanceLow, managed by AWSHigh, you manage everything

Swipe to see the full comparison ›

Real-world example: DoorDash uses a SageMaker model to detect fraud risk during customer claims, working alongside a generative AI bot that gathers call information first.


Our Recommended Path

The hybrid architecture that actually wins

The smartest approach for most organizations combines managed AI with custom intelligence, not an either-or choice.

  1. 1

    Lex or Amazon Q Handles the Conversation

    Natural conversation, FAQs, and basic troubleshooting, the layer that actually talks with the customer.

  2. 2

    Bedrock Keeps Answers Factual

    A Knowledge Base grounds every response in your real content through RAG, so nothing gets invented.

  3. 3

    SageMaker Handles the Specialized Work

    Fraud scoring, call summarization, and other custom tasks, wired in through Lambda where it actually adds value.

  4. 4

    A Human Takes Over When Needed

    If the bot can't resolve it, a live agent gets the handoff, along with the AI-generated conversation summary.

Talk It Through First

Not sure which architecture is right for you?

Every AI Fit Check starts with your actual call drivers and data, not a preference for one AWS service over another. We'll tell you honestly whether you need Bedrock, SageMaker, or both.

Request an AI Fit Check

Proof, Not Hype

We build this, we don't just diagram it

How this actually ships, not a theoretical architecture slide.

Real Build

Our own AI receptionist runs on Bedrock and Lex

We built an automated attendant using Amazon Bedrock and Claude, indexed against our own website content as a Bedrock Knowledge Base, with Amazon Lex handling the conversational layer. Real deployment, not a proof of concept.

Real Guidance

We recommend the architecture that fits, not the one that costs more

Bedrock and a Knowledge Base solves most self-service and agent-assist needs without touching SageMaker at all. We'll tell you when you actually need the more expensive path, and when you don't.

Certifications

AWS Certified Generative AI Developer – Professional badge
AWS Certified Machine Learning Engineer – Associate badge
AWS Certified AI Practitioner – Foundational badge

“From watching his YouTube videos, we engaged Peter on numerous contact center projects, including migrating a portion of our Cisco UCCX contact center into AWS.”

Amir SafayanReal Estate Entrepreneur

“We wasted a month trying to DIY Connect. DrVoIP had us live in 4 days, with full IVR.”

CX ManagerSaaS Helpdesk

“I feel their knowledge of both the AWS product as well as our previous product has been a great asset in easing the transition, as they already know how our current solution works and what we will want from AWS in regards to mimicking certain features.”

Jason AgcaoiliManager, ASMR

Where This Shows Up In What We Build

This stack powers services already on this site

You won't find “Bedrock vs. SageMaker” sold as its own line item. Here's exactly where it lives.

Knowledge AI (RAG)

Bedrock and a Knowledge Base is literally the engine behind grounded, document-based answers.

Explore ›

Agentic Automation

Where SageMaker's specialized models and Lambda-triggered workflows come into play.

Explore ›

Why AI Solutions

See the full picture of how DrVoIP applies AI across a contact center, not just this one piece.

Explore ›


Common Questions

Frequently asked questions

No. Most deployments start with just Lex and Bedrock. SageMaker only earns its place when you need a genuinely custom model, like fraud scoring or a domain-specific classifier, something Bedrock's general-purpose models can't do.

Retrieval-Augmented Generation. Instead of letting an AI model guess an answer from its training data, RAG has it look up the answer in your actual documents first, then respond using that content. It's the same approach behind our Knowledge AI (RAG) page.

When the task is a genuinely custom prediction, not a question. Fraud risk scoring, sentiment classification tuned to your business, or call outcome prediction are SageMaker territory. Answering “what's your return policy” is not.

No, though it overlaps with both. Amazon Q is a newer generative AI assistant that plugs directly into Connect, and it's often paired with the same Bedrock Knowledge Base underneath Lex-based bots.

Start with an AI Fit Check. We look at your actual call drivers, existing knowledge content, and whether you have a genuinely custom prediction problem, then recommend the simplest architecture that solves it.


Further Reading

More on AI, Bedrock, and Amazon Connect

AWS first LangGraph and CrewAI comparison diagram

AWS Bedrock, LangGraph, or CrewAI? Choosing an AI Stack That Survives Production

, , ,
Executive summary: Amazon Bedrock, LangGraph, and CrewAI are not three competing versions of the same product. Amazon Bedrock supplies managed access to AI models and supporting services. Frameworks such as Strands Agents, LangGraph, and CrewAI…
Amazon Connect direct dial smartphone illustration

Build an ai Receptionist for your call center

, , ,
An ai Receptionist? Back when vacuum tubes were still part of the computer science curriculum in most colleges, I read a book by Norber Wiener entitled "the human use of human beings".   As the title suggested, lets free humans to do the…
AWS search services options illustration

The DrVoIP AWS "Search Solutions" Guide

, , ,
which search service should I use One of the most common questions we hear from clients building on AWS especially those exploring generative AI is which search service should I use? AWS offers four distinct search and query services, each…
Helpful tips letter tiles

DrVoIP Amazon Connect Tech Tip - LEX Bot Versions!

, ,
Well, it is the 21st century and though we still drag around fax machines, we do seem to be getting away from Touch Tone Call Tree IVR systems!  Really, are you not tired of "Press 1 for this and Press 2 for that"?   I know I am at every…

A practical comparison of AI orchestration frameworks for teams building past the proof of concept stage.

How an AI powered receptionist gets built directly inside an Amazon Connect contact flow, using Bedrock and Lex.

A guide to the AWS services that make enterprise search and knowledge retrieval practical to deploy.

A practical look at managing Lex bot versions as a conversational AI flow matures past its first release.

AWS Bedrock, LangGraph, or CrewAI? Choosing an AI Stack That Survives Production

Build an AI Receptionist for Your Call Center

The DrVoIP AWS Search Solutions Guide

Tech Tip: Understanding Lex Bot Versions and Alias

AI Solutions

AI Solutions

Cloud Services

Conversational AI

Visit the Blog

Ready to design the right AI architecture for your contact center?

Tell us your call drivers and existing systems, and we will recommend the simplest stack that actually solves it, Bedrock, SageMaker, or both.

Request an AI Fit Check
844-4DrVoIP (844-437-8647)grace@drvoip.com

FREE GUIDES & CHECKLISTS

Take a useful next step.

Free resource 15 pages · PDF

Conversational AI - 2026 Edition

Fifteen-page visual guide to Amazon Connect AI agents, flows, Lex, prompts, guardrails, tool permissions, human handoff and architecture choices.

Get the Conversational AI Guide Delivered to your inbox
DrVoIP logo

Building cloud communications and contact center solutions since 2008.

DrVoIP on YouTube
Peter Buswell on LinkedIn

Services

Amazon Connect
AI Solutions
Cloud Services
Case Studies

Company

About Peter
Resources
Blog
Contact

Talk to us

844-4DrVoIP
grace@drvoip.com

© 2026 DrVoIP. All rights reserved.