Then you will want to check out the Strands Agents SDK. It’s a simple lightweight Python framework with a lot of powerful features.

They created a beautiful framework for setting up conversations, collecting the prompt, then under the hood it calls out to your LLM model of choice before responding.

It has guiderails, metrics, logs etc. to make the development process easier.

As you can imagine my favorite feature is the fact it supports Model Context Protocol so you can give the LLM Agents running in your application to a variety of insanely powerful tools.

It’s designed to deploy directly to AWS Lambda, AWS Fargate, Amazon EKS, Amazon EC2 right out of the box which is pretty handy.

I want to credit Jonathan Limbird (AKA JBird) with putting this one on my radar. If you would like to hear a deeper dive on this tech let me know and I bet I could convince Jonathan to do a Tech Talk on it.