The Blog
Learn something new with our experts
Running a System One Model of Your Own on Amazon Bedrock
Outside of chat, a lot of what we use language models for in production comes down to picking one of a few fixed answers. Is this support ticket urgent? Does this transaction look like fraud? Is this
Who Approved This? Delegated Authorization for AI Agents with AgentCore Identity
In 2019, a misconfigured firewall at Capital One was tricked into lending out its AWS credentials. The trick matters less than what those credentials could do: read about a hundred million credit appl
Building a Centaur Chess App with AgentCore Runtime and Strands Agents
In 1998, Garry Kasparov proposed a new format: human players paired with computer assistance. He called it "Advanced Chess," though the community settled on a better name: centaur chess. The idea was
Safe AI Development with Claude Code Permission Controls
Introduction You've seen how generative AI can automate boilerplate setup and accelerate project work to 80-90% completion, leaving you to focus on the complex, high-value tasks. The potential productivity gains are significant, and the use
Supercharge yourself with your own team of agents
Since entering the world of creating agents, one of the most powerful patterns being used is the supervisor, also commonly referred to as the orchestrator pattern. Very quickly explained: the supervisor pattern consists of a lead agent who
Let's Build an Agent on AWS!
AWS is going all-in on AI, and they're making it easier than ever to build agents in their ecosystem. They've recently released two key pieces: Amazon Bedrock AgentCore (a managed runtime for deploying agents) and Strands (an open-source Py
AWS AI Tooling: The Cheat Sheet Nobody Asked For (But Everyone Needs)
TL;DR: AWS offers numerous AI services, making it challenging to keep them all straight. This is your no-nonsense reference for actually figuring out which one solves your problem, without having to read a near-unlimited number of pages of
Robots Are Taking Our Jobs: Understanding AI Agents
TLDR: Agents are AI that don't stop at giving advice - they actually do the work. They try things, use tools, check results, and keep going until done. The gap between 'AI assistant' and 'AI colleague' is closing fast. From Human Loop to AI
Stop Copy-Pasting Into ChatGPT: How MCP Servers Actually Solve the Context Problem
TLDR: MCP servers let your AI agents directly access your company's data and tools. No more copy-paste, no more context switching. Just ask questions, get answers that actually know your business. Here's how to actually use them. If you hav
Moving Past the AI Hype (Introducing MCP)
In this post, I’ll cut through the noise surrounding Generative AI and the Model Context Protocol (MCP), explain the practical changes on the horizon, and show you why understanding these topics matters if you want to stay relevant in the i
Radical Simplicity in Cloud Architecture
For a long time, both professionally with clients in my consulting work and in the AWS community, I've been evangelizing the "serverless" paradigm shift. To me, it's crystal clear that you should focus on building what differentiates your b
The honeymoon is over
So our first year as a company went well. Great actually. Fantastic numbers, high spirits, and an everlasting startup feeling. That all changed early in year two. Act and react I already wrote some words in my “first-year review” regarding
How I built a Multiplayer App in 3 days
I thought it was time to create a follow up of my previous post Get your idea deployed to prod already. In the post I detail how you can quickly bootstrap an app and get it deployed and ready in no-time. This is a loose retelling of how I f
A great place to quit
Ok the title might be alarming but hear me out. There are a plethora of different awards for “Swedens best employer” or best/good/great place to work etc. You could argue that they are a bit watered down because of the amount of them. Or at
Shared CloudFront Distribution Cache Policy with SST
If you're sitting in a growing team and working with SST it's highly likely that you're working with personal staging environments. You might even use PR-deployments to test your sites before merging (smart of you). You might also have run
Testing Individual StepFunction States
AWS StepFunctions is a fantastic service for orchestrating complex workflows. But testing StepFunctions has always been... tricky. You could execute the entire state machine, observing side effects and final outcomes, but this approach ofte
Daddy, look what I made
I thought I’d mix it up a bit and add another flavour to the otherwise technical content that is presented here. This blog post will focus on my experiences and thoughts surrounding the first year at Elva, from the excitement of starting a
Building an IoT-powered Fridge
Imagine the scenario: guests are about to arrive, and you realize your fridge is running low on beer. Your BoA (Beverage on Arrival) SLA is in danger, and panic sets in, but fear not! In this blog post, we'll walk you through a solution we'
Get your idea deployed to prod already
You know that nifty idea you've been thinking about? The website that would revolutionize how you rule a small country/go grocery shopping? Whatever your idea is you can't share it with anyone because it's still in your head. It's time to d
Unlocking Scalability: Harnessing the Power of SQS and Lambda Integration
AWS Simple Queue Service (SQS) provides a reliable and scalable messaging solution, while AWS Lambda offers serverless computing capabilities. Combining these two services can enable robust and event-driven architectures. In this blog post,