Interview: “The coding aspect is just one small piece of the puzzle. There is the designing, coding, deployment, maintenance of the software and then operations,” says Srini Iragavarapu, AWS director of generative AI Applications and developer experiences. The AWS exec was talking to Dev Class about Amazon Q.

Iragavarapu goes on to suggest that ops teams from other platforms might also make use of Amazon Q, which was originally a coding assistant but is now an agent for software development. Its scope now goes beyond AWS services, although it is most suited to Amazon’s platform. It was announced in November 2023.

Iragavarapu makes the point that writing code is only one part of software development. “The coding aspect is just one small piece of the puzzle. There is the designing, coding, deployment, maintenance of the software and then operations,” he says.

Amazon Q is a brand that the cloud giant applies to various services, with Q Developer being the part that applies to software construction. The product began as Code Whisperer back in June 2022, when it seemed like a hasty response to GitHub’s CoPilot and one that was mainly adapted to coding for AWS services, having been trained in part on “internal Amazon repositories, API documentation, and forums.”

A lot has happened in the ensuing two and half years, including a rebrand to Q Developer earlier this year, the impact of Amazon’s “strategic collaboration” with Anthropic, and the steady addition of new features.

Last week AWS introduced inline chat support, where the AI assistant responds in the code editor rather than in a chat panel.

“Developers … didn’t want to switch context,” says Iragavarapu.”Not only do you get autocomplete on steroids, you actually get longer completions, you can optimize your code, add documentation … we’re using the Anthropic Sonnet 3.5 model.”

What about the perception that Q Developer is mainly useful for those coding applications that run on AWS and use its services? 

“If you are … working on AWS services, Q is enabled in the management console for AWS … you can ask questions about your resources, about your compute, about what you want to be able to do, that is a unique differentiator,” Iragavarapu tells Dev Class.

However, Q Developer is not just trained on AWS code but uses many other resources, he adds. The company has talked about migrating Java code JDK 8 to JDK 17, for example, and there “we use something called Open Rewrite,” though this is alongside other models. “They are not specifically trained on AWS information, they are generic,” Iragavarapu says. “There is also a feature called customization where a developer or an enterprise can ingest their code into their own environment and start getting recommendations tuned to their code.”

Features of Q Developer now also include multi-file edits, unit test implementation, planning and implementing new features based on prompts, and security scanning.

The market for AI developer assistants is crowded and there are a variety of pricing models. In the case of Q Developer, “there is a free version and I have my own friends ping me every now and then and ask me what’s the catch is? Actually there isn’t a catch,” Iragavarapu claims, though we note that there are (as one might expect) limitations, including no more than 50 interactions per month, and no more than 5 Q Developer Agent tasks. The paid version has limits too, though higher, and for tasks like code transformation, beyond a certain limit there is a cost of $0.003 per line of code submitted for transformation.

What’s coming next?

AWS tends to keep tight-lipped about future plans, but Iragavarapu did reveal that support for the Eclipse IDE is on the way, alongside the existing add-ons for Visual Studio Code, Visual Studio, and JetBrains IDEs. “There are some customers who are wanting Eclipse support. That is something that we’re working on,” he says.

There is also a general sense that “the abstraction layers are rising,” Iragavarapu adds, with AI able to implement and deploy entire features or even applications, rather than just helping with coding issues. In the case of Amazon Q Developer, it will likely deploy on AWS which makes this a strategic product for the company.