The Era of Production-Ready Agentic AI
If I had to sum up AWS re:Invent 2025 in one phrase, it would be this: The toy phase is over.
After a week of navigating 60,000+ attendees and clocking my standard 30,000 steps a day across the Las Vegas strip, the message from Amazon was unmistakable. We are no longer just “chatting” with AI. We are building the substrate for Agentic AI; autonomous systems that can reason, plan, and execute work across the enterprise.
This wasn’t just about new models but the operational backbone required to graduate agents from fragile prototypes to mission-critical infrastructure. A massive thank you to the AWS team for always hosting such a productive, informative event! They certainly gave us plenty to chew on this year.
Agentic AI as a Full Stack
The headline story of the conference was AWS’s decisive move to treat agents not as a feature, but as a platform. With the debut of Amazon Bedrock AgentCore, AWS defined a new blueprint for production-ready systems, providing the essential memory, identity, and tooling layers developers have been struggling to build from scratch.
This “full-stack” vision was further cemented by Nova Forge, which signals that AWS intends to own the “operating system” for autonomous agents. As we explored in our analysis, Agentic AI is becoming a full stack, shifting the focus from simple text generation to complex, multi-step problem solving. To support this, AWS also previewed an AI-native cloud built on frontier agents, showcasing how deep infrastructure integration will drive the next wave of innovation.
Multi-Cloud, Sovereign, and Governed
Agents need data, and that data lives everywhere. A major theme this year was AWS’s pragmatic embrace of multi-cloud connectivity. We saw this clearly as AWS championed an AI-first strategy that openly acknowledged the need for cross-cloud innovation. This was reinforced at the Analyst Summit, where the future of cloud was framed not as a walled garden, but as an interconnected ecosystem.
The ecosystem partners were ready to meet this moment. Nutanix launched GC2 for AWS GovCloud, enabling secure, air-gapped clusters for mission-critical workloads. Tessell announced expanded multi-cloud RDS capabilities to unify database management across environments, while Kong and AWS teamed up to advance AI governance, ensuring that as agents scale, policy enforcement scales with them. Even in the open-source world, we saw how AWS’s pragmatic strategy is addressing the AI trust crisis, balancing innovation with security.
Redefining Developer Productivity
One of the most insightful briefings I attended was on Amazon’s internal framework for measuring developer productivity. The shift from measuring individual output to “system-level” flow is a game-changer, and we are seeing this philosophy trickle down into the tools developers use every day.
For example, LocalStack hit record adoption as teams embrace local-first development to speed up iteration loops. We also saw how architectural modernization is meeting AI-assisted development, with tools helping teams refactor legacy code for the AI era. On the deployment side, Akuity expanded GitOps delivery to support multi-cloud targets, ensuring that modern delivery pipelines can reach wherever the application lives.
The End of Reactive Observability
If agents are autonomous, how do you know what they are doing? The answer lies in a new class of observability. Dynatrace made a splash by bringing full observability to AWS Agentic AI, allowing teams to trace reasoning paths, not just API calls. Similarly, our coverage highlighted how the era of reactive observability is over, with platforms pivoting from “fixing outages” to proactively predicting them before they impact the business.
Final Thoughts
Beyond the tech, re:Invent is always about the people. One of my personal highlights was getting to run a 5K with a fellow intern from my Intel days. Reconnecting 20-some years later is exactly what these events are about.
It was great to see that Mariah Carey’s performance was a big hit, and I had an incredible dinner at Best Friend in Park MGM. If you haven’t been, add it to your 2026 list. It lived up to the name, even if the owner isn’t actually my best friend (yet).
The shift to Agentic AI is going to keep us all busy for the next year, and I’m already looking forward to seeing what gets built with these new tools.
See you in 2026!
