What Just Happened | Justin Honaman, Global Head of Retail, Restaurants & Consumer Goods at AWS

For retailers looking to get more of an understanding of Agentic Commerce, this discussion lays it out. AWS is quickly announcing new products and if you are finding it hard to keep up, tune in. Not only will the roadmap be laid out, the strategy of what to do when you hit a road bump is discussed.

Since re:Invent in December 2025, AWS has announced Amazon Nova 2, Amazon Bedrock AgentCore, Amazon Quick, the expanded OpenAI partnership bringing GPT-5.5 and GPT-5.4 to Bedrock, Nova Act, and Kiro Autonomous Agent.

Each of these announcements represents a new frontier in what is possible for retailers and brands trying to build a better customer experience. But before any of it works, the foundation has to be right. And that foundation is data.

Honaman is direct about this. Most legacy retailers and consumer goods brands are not in a good place with their data. E-commerce data sits separate from the physical business. Systems do not talk to each other. Basic reporting is harder than it should be, let alone powering AI on top of it. The good news, he says, is that the technology is moving fast enough now to connect across data platforms and even across clouds, without requiring everything to live in one place. That dream of a single monolithic data platform is over. What matters now is access, cleanliness, and connectivity. Honaman points to platforms like Redshift, Aurora, Databricks, and Snowflake as the enterprise level tools driving this shift, and he is seeing a significant acceleration in companies moving to cloud specifically to get their data in a position where AI can actually use it.

Honaman then gets into Amazon Bedrock Agent Core, the platform that makes all of this governable at scale. This is not a separate purchase or a new subscription. If you are an AWS customer, you already have access to it. What it gives you is the guardrails, the rules of the road for how your agents operate, and the ability to move agents from development into production faster through composability. For those unfamiliar, composability means you can swap out and modify different parts of the platform without having to rebuild the whole thing. For retailers who are worried about committing to a build that could be obsolete in six months, that matters a lot. Honaman is clear that the platform is not in testing. It is live, customers are on it, and workloads are growing fast.

On trust and security, Honaman takes it back to 2006 when AWS launched and one of its founding principles was that your data is yours. Nobody at Amazon can access it. Nobody in your company can access it unless you grant them permission. That promise has not changed and everything being built on top of AWS today, including all of the agentic capabilities, carries that same security foundation. For retailers nervous about what it means to hand customer data to an AI platform, this is the answer.

Then there is the room itself. Honaman describes what is happening at AWS executive briefing centers in Seattle, New York, London, and Singapore. Every single agenda right now has AI on it. But what has shifted is who is in the room. It is no longer just the CIOs and CTOs. The CMOs, CFOs, and CEOs are showing up and they are not just listening. They are having near-technical conversations about what AI can do for their business. Honaman says a year ago that was not possible. Today it is the expectation.

Buy For Me is one of the most concrete examples of agentic commerce that Honaman walks through. If a shopper searches for a product on Amazon that is not sold on Amazon, the agent goes and finds it, navigates to the brand's website, and can complete the purchase autonomously. It is powered by Amazon Nova, Claude models from Anthropic, and runs on Amazon Bedrock. It is not a concept. It is live.

Honaman wanted to go deeper on supply chain and that is a conversation for next time.