#47 AI Agent Kills Production 😈

This week, our serverless expert is AWS Community Builder Jon Goodall, our spotlight falls on AWS Serverless Hero AJ Stuyvenberg, we look at the latest service releases, news, articles, & more!

Jul 27, 2025

Welcome

In last week’s issue, our serverless expert was AWS Community Builder Seth Orell, and our spotlight fell on AWS Senior Solutions Architect Urmila Raju!
This week, our serverless expert is AWS Community Builder Jon Goodall, our spotlight falls on AWS Serverless Hero AJ Stuyvenberg, we look at the latest AWS service releases, blog posts, hints and tips, news and more!
This week’s newsletter is sponsored by Leighton.

A Glimpse into My Week 🎤

This week, I published an article titled “Amazon Aurora DSQL Sidecar to DynamoDB — Part 1“:

This is a seriously useful pattern in my opinion, especially where you want to extend existing solutions for new access patterns that would not be possible with DynamoDB, and you don’t fancy a large-scale migration from DynamoDB to DSQL! Go check it out!

📰 Articles that caught the eye

Here are some stand-out articles I read during the week in the World of Serverless, AI, Engineering and Architecture!

⭐ My favourite this week was by David Behroozi, covering how to make your cold starts quicker!

Marko from Serverless Life covers “Lambda Live Debugger vs Official AWS Remote Debugging“.
InfoWorld covers ‘AWS imposes caps on Kiro usage, introduces waitlist for new users’.
Chris Dobson writes about ‘A local Lambda development environment’ in this article.
Carlos Filho talks through creating a new tool called ‘Anomaly Guardian’ in this detailed article.
Hector Zelaya has a great article titled ‘How to Build a Serverless Voice AI Assistant for Telephony in AWS using Twilio ConversationRelay‘.
David Behroozi discusses ‘The Fastest Node 22 Lambda Coldstart Configuration‘.
Joseph Ndambombi Honpah has an interesting solution called EchoGuard, for analysing audio recordings for compliance using AI.
Lilupa Karu has a detailed article on ‘Enterprise Bedrock Agent for Retail Customer Success‘.
Talex Maxim has a very clickbaity article on ‘Serverless: The Illusion of Choice‘.

🎓 Ask the Expert

Each week, I ask a different serverless expert the same three questions to get their personal insights - this week, we have AWS Community Builder Jon Goodall:

Opinions are the author’s and do not express the views of their employer.

1. What is one common mistake you see teams making when implementing serverless solutions, and how can they avoid it?

Using the wrong service - or using a service without looking at all available options to solve the problem.

I've seen teams go directly for Lambda as a first option, but when they hit issues around something like cost predictability they jump fully into Fargate instead. Containerising functions is a massive development effort. You then lose some of the native functionality within Lambda - e.g. long polling of SQS queues - so have to build that yourself.

Conversely, I've seen teams stay with Lambda when the data suggests moving to Fargate would be a better fit. For instance a workload that cannot tolerate a cold start, and to resolve this they've put in provisioned concurrency, at massive cost. Whilst this does solve the problem, a far more cost-effective way to resolve is to move that workload to something that charges by the hour anyway, as the cost is much lower.

In short, match the service to the workload requirement, don't just use one thing for everything.

2. Which serverless tool or service are you most excited about right now, and why?

I've been a fan of StepFunctions for a very long time for the way they make orchestrating complicated workflows really simple.

Whilst I'm not a massive user of the SDK integrations between StepFunctions and other services, they can be a really powerful way of performing CRUD operations or sending notifications without having to write "glue" Lambdas.
That being said it makes testing a bit more involved, as you have to test the function end-to-end in an AWS account with data - but you should be doing that anyway so I don't see it as a downside.

Anything that reduces the amount of code I have to write is a good thing in my book.

3. What is your favourite trick or tip when working with serverless that the readers may find interesting?

I come from more of an Ops background, compared to a developer, so I previously struggled with "how do I even get this thing to deploy" and "how does the data move around this thing", so I start there in all new projects now.

Start from the IaC side of things using whichever tool you like best - personally I use AWS SAM as I think it's the most "put together" offering, plus it extends CloudFormation, so if there's something else that you need in the stack that isn't serverless, you don't have to worry about writing something else to deploy it.
Once you've got a your functions and workflow stubs deployed, fill in from there. I find my brain works much better "outside in" in this way.

Also, if you're used to using a specific IDE for your dev work, maybe try a different one. They don't always have feature parity between the AWS extensions - e.g. VSCode can dynamically draw Step Functions State Machines in the IDE, but I find the auto-complete for IaC much better in the JetBrains offerings.

✅ Bonus tip: join the hashtag#believeinsls discord! There is a community there to answer any questions you may have without getting overzealous on serverless or without judgment! Check it out!

🧠 Tips & Tricks

This week’s tip is around SNS Subscription Filter Policies, and a lesser-known complexity that people often forget about when filtering on the payload body over message attributes, but it can bite them as their app grows.

If you are filtering on the body of the payload, be mindful that the nested nature of an event or message, and the number of array properties in each leaf node you are filtering on, all contribute to a hard limit of 150:

“The total combination of values in a filter policy must not exceed 150. To calculate the total combination, multiply the number of values in each array in the filter policy.“

This means, for an example filter payload of this:

{
  "detail": {
    "data": {
      "company": ["CompanyA", "CompanyB"]
    },
    "metadata": {
      "source": ["CRM"],
      "type": ["PersonCreated"]
    }
  }
}

You can only ever have a maximum of 5 companies in the company array (or equivalent that you might filter on in your policy). Go for filtering on message attributes instead.

🚀 New Releases

Here are the latest and most interesting releases this week in the AWS World:

⭐ This week was very quiet, but my favourite was the new SQS fair queues for multi-tenant workloads - very cool!

🔥 Tip: Check out https://aws-news.com/ for the very latest up-to-date serverless releases as they happen, created by the talented AWS Serverless Hero Luc van Donkersgoed.

👷🏻 Tools & Frameworks

Check out the latest open-source frameworks, news, and tool updates from the past week.

rulesync - A Node.js CLI tool that automatically generates configuration files for various AI development tools from unified AI rule files (.rulesync/*.md). Also imports existing AI tool configurations into the unified format.
Event Catalog - You can now embed Entity Maps directly into your documentation with Event Catalog, giving your team instant visibility into the core concepts of your system—like Orders, Users, or Products—right where they need it.
CloudPouch.dev - releases version 1.42.0 with New Cost Insight: Amazon EKS Extended Support Detection.

✖️ Social of the Week

This week’s social is by Serverless Hero Yan Cui, discussing the injection of malicious prompts into the Amazon Q's VSCode extension.

Well, this was bound to happen at some point with malicious prompts, but in an AWS open-source tool? Yikes! I’m super interested in viewing the postmortem on this one personally and seeing the lessons learned.

What are your thoughts on this? Feel free to leave a comment below.

🎙️ YouTube & Podcasts

Here are some of my favourite videos and podcasts this week.

⭐ My favourite video this week is by the Primeagen, talking through a real-life disaster as someone’s coding agent went rogue and deleted the production database - very fun!

Sedna and Ran Isenberg talk about ‘Go to Market with MCP‘ on the Let's Build a Startup channel.
James Eastham talked about ‘Thinking in Events: Principles of Event-Driven Architecture, Part 1’ on the Compiled Conversations podcast.
Derek Comartin covers ‘Why Event-Driven Architecture Feels Broken‘.
The Primeagen has a fun video on ‘AI “Destroys” Months of Work‘.
Dr Nicole Forsgren's CONTROVERSIAL Take On Software Engineering vs. Craftsmanship on the Modern Software Engineering channel.
Alan Smith covers ‘Inside GPT – The Maths Behind the Magic‘ on GOTO conferences.
The latest episode titled ‘Penna the tiny, unobtrusive, structured logger for Java’ on Java In The Cloud.

Weekly Case Study 🔍

This week’s case study comes from KONE:

KONE Corporation, a global leader in elevators and escalators, implemented a generative AI solution using Amazon Bedrock to enhance their field service operations. Their primary use case involves a technician assistant that helps over 40,000 field technicians per day to resolve equipment issues more efficiently, handling approximately 30,000 technical help desk calls monthly.

🗣️ Inspirational Quotes and Thoughts

This week’s inspirational quote or law is the law of parsimony (Occam's razor) by the 14th-century English philosopher, William of Ockham:

“The simplest solution is most likely to be correct.“
Occam's razor - William of Ockham

Who thought a 14th-century Friar would say such a fantastic quote, which many years later would be attributed to software design! I love this quote, and this is one of the principles I have listed in the way we design software in the AWS Practice at Leighton.

In my experience, starting with the most simplistic solution to a problem is often the right direction as:

Chances are, the requirements might change over time, and so will your solution.
You will add features that will never be used.
You want to get initial feedback from the customer and iterate on it.

What are your own thoughts and experiences of this quote? Feel free to leave a comment below.

🗳️ Poll of the Week

In last week’s poll, we asked the question “Does your organisation have reusable base components, frameworks and patterns to use on projects?”.

Interestingly, 67% of people said their companies are doing some of the list, and 33% said yes, all of the above. It’s great to see so many companies focusing on reuse.

This week, we ask the question, “Are you comfortable working with multiple languages in your AWS solutions?”. Do you typically just work with one programming language day to day, are you comfortable with many, or somewhere in between?

Feel free to leave a comment below on why you chose your answer and your experiences!

📅 Serverless Events

The following serverless events are upcoming, so mark your calendars.

🎟️ To note, CFP is currently open for the AWS North Community Conference, which I am helping organise, and we also have opportunities for lightning talks throughout the day. Go check it out!

Other fantastic events happening soon:

ACD Singapore - 2nd August 2025
GOTO Serverless Bengaluru - 8th August 2025
ACD Australia - 15th August 2025
ACD Adria - 5th Sept 2025
AWS Community Day Baltic - 10th Sept 2025
ACD Bay Area - 17th Sept 2025.
ACD Aotearoa - 18th Sept 2025
ACD Poland - 18th Sept 2025
ACD Portugal - 27th Sept 2025
ACD DACH - 7th Oct 2025
ACD Nordics - 10th Oct 2025
AWS North Community Conference - 16th Oct 2025
ServerlessDays Milano - 21st Oct 2025
ServerlessDays Cardiff - 23rd Oct 2025
Serverless Architecture Conference - 20th-22nd Oct 2025
ServerlessDays Sao Paulo - 5th Nov 2025

Do you have any upcoming events that you want to highlight? Message me below!

⭐ Spotlight

This week’s spotlight falls on AWS Hero AJ Stuyvenberg:

AJ is a very well known AWS Serverless Hero, international conference speaker, YouTube content creator, open-source contributor, podcast guest, and more; and is currently a Staff Engineer at Datadog, specialising in serverless observability and distributed system monitoring. He was also previously a Principal Software Engineer at the iconic Serverless Inc where it all started!

Thank you for everything you do for our amazing community AJ!

Thank you for reading the latest Serverless Advocate Newsletter!

If you want to find out a little more about me, please have a look at:

https://www.serverlessadvocate.com/

See you next time,

Lee

Twitter | LinkedIn | YouTube | BlueSky | Website