Intro
I’ve been using Amazon Q for command line for a few days building AWS IoT solutions…
Summary: It is one of the best AI tools for backend development I’ve ever tried!
More detailed thoughts and experiences
Amazon Q chat, used from the command line, fluently navigates between the local source code and existing AWS infrastructure, providing end-to-end visibility and suggesting tailored modifications. I used various AI tools for AWS IoT development and must admit that Q is clearly the best option based on my experience. I encountered several issues and had to provide precise instructions for Q, but overall, Q was a great assistant.
Most surprisingly, Q is able to help with IoT-specific tasks, which is often a challenge for other AI assistants.
Issues & Solutions
❌ Issue: By default, Amazon Q hardcodes values instead of using the existing configuration file from my project.
✅ Solution: Instructing it to use the config file once was enough; Q read and “understood” my config file and extended it with appropriate values.
❌ Issue: Sometimes, Amazon Q creates IAM Roles “by hand” instead of using CDK functionality.
✅ Solution: Instructing it to utilize full CDK capabilities solved this problem, and Q adequately updated my CDK App to manage permissions.
❌ Issue: Amazon Q is mixing the AWS CLI and CDK functionalities and suggests managing AWS resources using CLI when CDK is clearly a better option.
✅ Solution: Instructing it to manage “static” AWS assets via CDK and “dynamic” assets via boto3/CLI. This part is a bit tricky, and developers should pay attention to Q’s recommendations.
❌ Issue: Amazon Q omits profile when invoking CLI actions.
✅ Solution: At the very beginning of the conversation, inform Q about the CLI profile and default region it should use. I also strongly advise verifying the proposed CLI invocations BEFORE executing them.
❌ Issue: Amazon Q is having trouble responding right now; unhandled error (ThrottlingException).
✅ Solution: The Amazon Q free option lasted for 3 days of solution development - now it demands that I upgrade to Pro Tier. I studied the AWS pricing and documentation, looking for Pro Tier limits. Still, AWS does not provide specific quotas mentioning “high limits of advanced features” (please correct me if I’m missing something).
My Technical Stack
I manage “static” AWS infrastructure using Python and AWS CDK; for “dynamic” assets, I use Python and boto3.
I use Jupyter Notebooks to interact with AWS APIs and log into the AWS Web Console when absolutely necessary.
👉 Have you already tried Amazon Q for the command line? Please share your experiences!
🥸 Disclaimer: This is NOT an AWS advertisement but my actual experience in working with this tool.