Title: Issue #241: Using AI to GSD (Get Stuff Done) Publication: Ready, Set, Cloud! Series: Ready, Set, Cloud Picks of the Week Author: Allen Helton Published: November 09, 2026 URL: https://www.readysetcloud.io/newsletter/241/ How confident are you that the generated code your agent wrote will work in production? This week's issue helps us figure out how to boost our confidence and get better quality results. ### 🦸 Community Superhero Our community superhero this week is [Daniel Balcárek](https://www.linkedin.com/in/daniel-balcarek/), senior engineer at Komerční banka and a wonderful open source developer. He's published some great Angular libraries, and has over 40 published posts on Dev.to. He also has some fun and clever ideas with his usage of AI, which are fun to read and learn about. Thank you for everything you do, Daniel! ### 💯 Spotlight We've seen plenty of content about getting AI to write code. But getting it to write code that matches your definition of done is a different story, and [Teri Radichel](https://www.linkedin.com/in/teriradichel/) did a wonderful job telling us how [she does it in her post from last week](https://teriradichel.substack.com/p/getting-ai-to-gsd). I love how honest she is about where things have gone wrong for her. My favorite (aka most terrifying) example was an agent granting IAM admins in every environment the ability to assume roles in every other environment. It's a long read, but I nodded along the whole way. No spoilers here, it's worth the read yourself. ### 🔥 My Favorite Content It can be a bit overwhelming honing your setup for agentic development. How do you know what to pick? And how to you make the tools you like even better? Stepping away from the hundreds of LinkedIn ads that claim to "solve all your problems," I found a Hacker News thread where people are [sharing their agentic setups](https://news.ycombinator.com/item?id=50366117) and offering advice and opinions on what works for them. I really like this because it shows you don't have to go crazy setting up tools because some of the basics get you pretty far on their own. I'd say we all have an agent or two (or ten) running in the cloud these days. And that's great, it's making us more productive than ever. But are you monitoring their decisions? Do you know *why* they did what they did? I'm sure not. Well, I haven't been. I published an article on [what the heck we alert on anymore](https://www.readysetcloud.io/blog/allen.helton/what-the-heck-do-i-alert-on-anymore/) with agents fundamentally changing the attack surface of our applications. It's a sponsored post that goes into some innovative ways for keeping AI agents honest so we don't have another Hugging Face incident like we did a couple of months ago. I'm a sucker for some deep, innovative engineering, and that's exactly what we got from [Karthik Raman](https://www.linkedin.com/in/karthik-raman-247b36104/) and [Brian Celenza](https://www.linkedin.com/in/bcelenza/) last week where they describe [GitHub's new git storage architecture](https://github.blog/engineering/architecture-optimization/inside-githubs-new-git-storage-architecture/). GitHub has been on a migration path to Azure this year, and part of this migration is updating the underlying system to adapt to the skyrocketing agent traffic. The article goes through the approaches they took in "Project Rubicon" and does a wonderful job explaining what they're doing to hit massive scale. Super cool stuff. And speaking of engineering for scale, AWS had a couple of pre:Invent releases last week for exactly that. We got a new [queue consumption mode for Kafka in Lambda ESMs](https://aws.amazon.com/blogs/compute/introducing-queue-consumption-mode-for-apache-kafka-in-aws-lambda/) that take advantage of Kafka Share Groups. Now you can add a `Stream` or `Queue` consumption mode to your existing ESM for Kafka to Lambda and you're off to the races. We also got [bulk put items for DynamoDB](https://aws.amazon.com/blogs/database/introducing-bulk-put-items-for-amazon-dynamodb-load-data-from-amazon-s3-into-existing-tables-at-scale) that lets you give DDB a location in S3 for your input files, and it loads the content as an async job. Best part? It doesn't compete with your read and write capacity for the table. ### 💡 Tip of the Week I'm sure we've all seen that mysterious line of code that looks harmless but when you remove it everything starts failing. Or seen that server that when it reboots your CI pipelines all go red for a week. With AI doing most of our coding (and reviewing) these days, I wonder how many more of these types of anomalies are going to start popping up in the months and years to come. [Adrian Hornsby](https://www.linkedin.com/in/adhorn/) shared a comic last week that hits the nail on the head. ### Last Words On a personal note, I starting a new job today! I've joined [Zillow](https://www.zillow.com/) as a Principal Engineer to help run the Transaction AI org. I'm super excited to be a part of the innovations coming to the real estate industry, especially with creative uses for AI. It's going to be fun 😁 What did I miss? What made you nod along (or 🙄)? Hit reply if you're reading the email. Prefer socials? Ping me on [Twitter](https://twitter.com/allenheltondev), [LinkedIn](https://www.linkedin.com/in/allenheltondev/), or [email](mailto:allenheltondev@gmail.com). Happy coding! Allen