IBM, AWS unite to scale trustworthy AI with seamless governance integration

Keith Shaw: Hi everybody, welcome to DEMO, the show where companies come in and show us their latest platforms and products. Today, I'm joined by Neil Leblanc, the Go-To-Market Lead for Watsonx.governance at IBM, and Eduardo Fronza, a Partner Solutions Architect at AWS. Welcome to the show, gentlemen.

Neil Leblanc / Eduardo Fronza: Thank you. Keith: All right, so what are we seeing on DEMO today? You've got a couple of products, right? Neil: We do.

What we're excited to share with you — and with the audience — is how we're bringing governance and data science together to provide an offering that helps organizations scale their AI initiatives.

Keith: Okay, is it one product from the IBM side and one from the AWS side? Or is it a partner product? Neil: Great question. It's Watsonx.governance from IBM and Amazon SageMaker AI from AWS.

We’ve built an integration — not necessarily a unified product — but a seamless experience that delivers value for customers. Keith: And within a company, there would be a number of different roles this integration is geared toward. Who's the main audience for this product?

Eduardo: On the Amazon SageMaker and AI side, the primary users are ML engineers, data scientists, and project managers working in the AI/ML space.

Eduardo: What’s really nice about this integration is that these personas — across SageMaker and Watsonx — don’t need to change how they already work with these services. It’s a seamless integration designed to help them derive more value and insights from the tools they’re already using.

Keith: And Neil, on the Watsonx side, are there other groups that benefit? Neil: Absolutely. And Eduardo, I owe you an apology — it should be Amazon SageMaker, not AWS SageMaker!

But yes, complementing the data science side, the governance layer engages business users, including CXOs, chief privacy officers, chief data officers, and chief information officers. They collaborate to define expectations and outcomes when leveraging AI.

Eduardo: To build on that, the integration leverages SageMaker's model registry — a construct where ML teams register models along with training metadata and evaluation metrics. That metadata is automatically populated in Watsonx.governance. Any changes made on the governance side are reflected back in SageMaker.

This provides full lifecycle visibility and collaboration across technical and business roles.

Keith: So let’s ask the core question we pose to all our guests: What problem does this solve? Or in other words, why should companies care? Neil: There are a few key problems — most notably, trust. Organizations want to ensure that their AI deployments are secure, compliant, and trustworthy.

It’s not just about the integration; it’s about the capabilities this integration enables — so companies can move from proof-of-concept into actual production-level AI deployment.

Keith: And without this integration, what are companies doing instead? Neil: All of the above — manually gathering approvals, emailing stakeholders, struggling to capture metadata about models and assets. It becomes incredibly labor-intensive. Automating governance — and mapping it to regulatory obligations — makes the whole process more efficient.

A joint solution like this helps organizations stay ahead of complex rules and compliance demands.

Keith: All right, let’s jump into the demo. Eduardo, I’ll let you take over. Eduardo: What we’re seeing here is the Watsonx.governance dashboard. It’s essentially the cockpit or landing page for managing AI across the organization. It's fully configurable.

In this example, the use case is detecting insurance fraud using AI. Watsonx.governance includes a questionnaire — here, 50 questions — that helps assess risk posture. Based on answers, it identifies risk levels using the IBM Risk Atlas. For instance, one risk here is data bias.

You can capture risk level, mitigation strategies, and more—then roll it all up to a use case and initiate a workflow.

Eduardo: Through that workflow, the use case is routed to different stakeholders for review and approval. Once approved, Watsonx.governance automatically creates a "model group," which triggers the SageMaker integration.

Eduardo: From there, the data science team starts building and training the model. They register it in the SageMaker model registry. All training metadata is stored, then reflected back in Watsonx.governance.

This allows compliance and risk officers to review and decide whether the model is ready to advance in the lifecycle.

Eduardo: Once the model is approved within Watsonx.governance, that approval flows back into SageMaker. If the model isn't approved, it won’t be deployable through the ML pipelines. Once it is, deployment to test, QA, or production can proceed — and Watsonx.governance captures the model’s new lifecycle stage.

Keith: That’s it, right? Eduardo: That’s it! Keith: Great. So how can customers get started? Do they need to be clients of both AWS and IBM? Eduardo: Watsonx.governance is available on the AWS Marketplace. From a governance perspective, customers can explore and get started directly there. Neil: Exactly.

SageMaker is a fully managed service on AWS, so if customers already have an AWS account, they can begin training models right away.

Watsonx.governance is available as both a fully managed SaaS offering from IBM and as a deployable option on a customer’s own Amazon VPC for those who need more control.

Keith: Great stuff, gentlemen. Thank you again for joining us and showing off this fantastic integration. Neil / Eduardo: Thank you! Keith: That’s going to do it for this week’s show. Be sure to like the video, subscribe to the channel, and leave your thoughts in the comments.

Join us every week for new episodes of DEMO. I’m Keith Shaw, thanks for watching!