My thoughts on Swami’s AWS Summit announcements

Dr. Ryan Ries here, and I'm writing this from my hotel in NYC after an absolutely packed AWS Summit yesterday.

Swami Sivasubramanian took the stage as AWS VP of Agentic AI, and let me tell you - his focus was crystal clear: production-ready AI agents at scale. AWS is doubling down on solving the hardest parts of getting agents into real business workflows.

Quick sidebar: if you were at my breakout session yesterday, thank you! I hope you got a lot of value out of it.

The Undifferentiated Heavy Lifting Problem

Getting agents to production is hard. The security concerns, the memory management, the tool integration complexity... it's all undifferentiated heavy lifting that keeps teams stuck in prototype hell.

That's exactly why the Amazon Bedrock AgentCore announcement got my attention. This service is AWS saying, "We'll handle the infrastructure headaches so you can focus on building agents that actually solve business problems."

The genius of AgentCore is that it's modular. You don't need to rip and replace your existing agent framework. Whether you're using CrewAI, LangGraph, LlamaIndex, or Strands Agents, AgentCore wraps around your code with enterprise-grade capabilities:

• AgentCore Runtime: Serverless environments with session isolation (no more data leakage between users)
• AgentCore Memory: Both short-term and long-term memory with semantic search capabilities
• AgentCore Identity: Proper OAuth integration with services like GitHub, Salesforce, and Slack
• AgentCore Gateway: Transforms your existing APIs into agent-ready tools via MCP protocol
• AgentCore Observability: Step-by-step visibility into what your agents are actually doing

AgentCore eliminates the "prototype valley of death" that's been plaguing AI implementations. You know, that gap between "cool demo" and "production system that doesn't break when 1,000 users hit it simultaneously." (We’ve all been there, I get it!)

At Mission, we built an agent using this system, and the difference in development velocity is night and day. Max Goff (Senior Solutions Architect, GenAI on my team) sent a message in our Slack that encapsulated my thoughts on AgentCore as well. “If 2025 is the Year of the Agent, then AWS is totally on target with their offerings.” A ton of people we spoke with were excited about AgentCore and eager to get started.

Nova Gets Serious About Customization

The big Nova news was customization through SageMaker AI. AWS finally launched extensive customization capabilities across all stages of model training - pre-training, fine-tuning, alignment recipes, the works.

This is a huge step forward. To make Nova more accurate, the AWS team launched these customizations specifically to help customers see the results they need from these models. No more "one-size-fits-all" limitations.

Max and I had some interesting conversations about Nova Act. In truth, we're still not 100% sure how it will compare to other agentic solutions. Time will tell.

The Agent Development Acceleration

AWS has been working hard on improving MCP (Model Control Protocol) to make agent development easier, and it shows. They outlined the 4 keys to successful agents:

1. Model & application capabilities
2. Ability to discover, publish, & procure
3. Services to deploy & operate
4. Tools & frameworks to build

They're pushing virtual try-on as a great way to blend products into environments. This is a good fit for many industries, such as retail and manufacturing, anywhere you need to visualize AI recommendations in real-world contexts.

The Surprises That Made Me Think

S3 Vectors was just announced, and honestly, I don't know what to even think about it yet. I guess you can use OpenSearch in a tiered fashion to optimize costs, but... data was never really the issue with vector storage. This feels like it’s solving a problem that wasn't the bottleneck to production for most companies.

Kiro also launched earlier this week. It's a coding IDE designed to take on Cursor and other AI-powered coding platforms. Interesting move into the developer tooling space.

What This Means for Your Business

AWS is clearly pushing to help companies build agents faster and more securely. This is exciting because it means companies can stop reinventing the wheel on agent infrastructure and start focusing on the business logic that actually matters.

The companies that understand this shift first will have a massive advantage. Instead of asking, "How do we add AI to our workflow? " They'll be asking, "How do we redesign our workflows to be AI-native from the ground up?" AWS is betting big that they can remove the technical barriers that have been holding teams back.

Have you started experimenting with agents in your organization yet? Let me know! I also would love to hear your thoughts on the keynote announcements. It’s an exciting time to be in this space.

Pictured: Me, Alexa Rubin (Sr. Customer Success Manager), Max Goff (Senior Solutions Architect, GenAI), and Olivia Martinez (Director of Partner Marketing) at Mission.

Until next time,
Ryan

Now, time for our AI-generated image and the prompt I used to create it.

Create an image of me in NYC at the Javits Center. It's busy, and you can see a lot of people in the background heading to the AWS Summit sessions. I am wearing a black Mission t-shirt, jeans, tennis shoes, and I have a backpack on. I have attached a reference photo of myself.

I wouldn’t say this looks much like me, but it did really nail the AWS branding.