🦸 Community Superhero

Our community superhero this week is Chloe Condon, senior Developer Advocate at Workato and creative technologist known for bringing joy, accessibility, and genuine community care into the engineering world. Chloe builds wonderfully strange and clever projects in public. She bridges hardware, AI, and creativity to use her platform to amplify new voices who are still finding their footing. She mentors beginners, supports career switchers, and consistently shows up for people behind the scenes in ways most never see. Thank you for making tech feel more welcoming, more creative, and more human, Chloe!

💯 Spotlight

DynamoDB just released multi-attribute composite keys for GSIs, which is a mouthful but an incredibly powerful feature. Jason Butz wrote an article last week explaining what it is and why it’s important. For a tl;dr, you don’t need to stick together attributes manually for your Global Secondary Indexes anymore - DynamoDB will do it for you! Jason does a great job covering everything you need to know and giving you a easy to follow explanation. It seems to me like this will help prevent some single-table design overuse in our systems (which is a good thing!).

🔥 My Favorite Content

Darryl Ruggles is back with a fantastic write-up on a serverless app he built for sales analytics with AI insights. The app is written in an architecture we’re all accustomed to, making it a familiar setup, but what I love about the article is the explanation of the data format he chose and what benefits he gains from doing it this way (no spoilers). I also like that he shows us AWS alternatives that do the same thing, but for a much higher cost. Super cool build and great explanation!

In a general post about software engineering, I published an article reminding developers that breakthroughs are just boring updates that stack up. In many instances, it often feels like the day to day grind is just a drop in the bucket and never makes a difference. But after a while, these updates stack up to breakthrough capabilities and major milestones. I walk through a case study with Valkey 9 how they made somewhat “boring” updates that all ended up making it run at a whopping 1 billion requests per second 🤯

I came across a great article from Austin Parker titled “Fast and Close to Right: How Accurate Should AI Agents Be?” where he argues that hallucinations aren’t actually the biggest problem for AI agents in observability. Austin claims the real challenges are managing data fidelity and task accuracy trade-offs in agentic systems. He poses if we’re wrong to try to force AI agents to be deterministic (treating every LLM quirk as a bug) and suggests a more nuanced approach to evaluating agent reliability. It’s a fascinating perspective that made me rethink how we measure accuracy for AI in complex systems.

💡 Tip of the Week

I had a hard time finding posts on social media that weren’t people traveling to Vegas for re:Invent. But I did find one that I found exciting - Kiro, the agentic IDE from AWS, went generally available! If you haven’t given spec-driven development a try, it’s absolutely worth a shot. I’ve been using it as my daily driver for a few months and love it. Congrats, Kiro team!

🐣 New Releases

As we know, releases have been happening every day in full force. To get a recap of everything that has come out, visit AWS News for summaries, filters and more. Here are a few of my favorite announcements.

Claude Opus 4.5 is now available in Bedrock. It’s available in Kiro as well!

Node.js 24 is now supported in Lambda. Rejoice!

You can now block public access for S3 buckets at an organization level. Great update for security posture.

Last Words

If you’re at re:Invent, come find me! I’ll be around the community lounges all week. I can’t wait to see everybody!

That’s my take on the week, but what’s yours?

What did I miss? What made you nod along (or 🙄)? Hit reply if you’re reading the email. Prefer socials? Ping me on Twitter, LinkedIn, or email.

Happy coding!

Allen