Amazon Renames CodeWhisperer, Adds New Functions

Amazon’s new Q Developer joins the Q Business family, aiding devs with coding and AWS tasks.

CodeWhisperer is now a member of the recently-announced Q Business and the Amazon Q family of business-oriented generative AI chatbots, of which Q Developer is a member. Like CodeWhisperer, Q Developer, accessible via AWS, assists developers with various daily duties, including app debugging and upgrading, troubleshooting, and security scanning. 

Doug Seven, general manager and director of AI developer experiences at AWS, suggested in an interview with TechCrunch that CodeWhisperer was a branding failure. Third-party metrics support this; despite offering a free tier, CodeWhisperer encountered difficulties matching the progress made by its primary competitor, GitHub Copilot. Copilot boasts more than 1.8 million paying individual users and tens of thousands of corporate clients. (Failurely initial impressions certainly did not help.) 

Seven stated, “CodeWhisperer was our initial foray into code generation; however, we aspired to establish a brand and name that could accommodate a broader range of use cases.” “Q Developer can be viewed as the progression of CodeWhisperer into a significantly more comprehensive platform.” 

Q Developer is capable of testing code generated with SQL, a programming language frequently employed for database creation and management, in addition to assisting in transforming and implementing new code conceived in response to developer queries. 

In the same way as Copilot, clients can optimize Q Developer for their internal codebases to increase the applicability of the tool’s programming suggestions. (Also, the deprecated CodeWhisperer provided this alternative.) In addition, Q Developer is equipped with the Agents capability, which enables it to perform tasks such as feature implementation, code documentation, and refactoring (i.e., restructuring) independently. 

When a user submits a request to Q Developer, such as “add to favorites,” in their application, Q Developer analyzes the application code, generates new code if required, devises a step-by-step plan, and finalizes code tests before implementing the suggested modifications. Before Q implements the plan, developers can review and refine it by connecting steps and applying revisions to the required files, code blocks, and test suites. 

“In the background, Q Developer creates a development environment in which the code is executed,” Seven explained. “Q Developer, therefore, takes the entire code repository, creates a branch from that repository, analyzes the repository, implements the requested changes, and returns those modifications to the developer in the case of feature development.”

Amazon also claims that agents can manage and automate code upgrading processes. Java conversions (specifically Java 8 and 11 built with Apache Maven to Java version 17) are going live today, and .NET conversions are forthcoming. Seven further stated, “The Q Developer examines the code for any areas that require improvement and implements those modifications before returning it to the developer for review and commit.” 

Agents resemble GitHub’s Copilot Workspace in that both facilitate generating and implementing plans for software problem fixes and new features. Similar to Workspace, I remain skeptical regarding the efficacy of this more autonomous methodology in resolving the challenges associated with AI-driven coding assistants.

According to a GitClear analysis of over 150 million lines of code committed to project repositories over the past several years, Copilot was increasing the amount of erroneous code submitted to codebases. Security researchers in another jurisdiction have cautioned that Copilot and comparable tools have the potential to magnify pre-existing vulnerabilities and security concerns within software development endeavors. 

This is not unexpected. Programming assistants fueled by AI appear impressive. However, due to their training on pre-existing code, their recommendations mirror recurring themes in other programmers—work that may contain significant flaws. The errors introduced by assistants’ educated estimates are frequently challenging to detect, mainly when developers who rapidly adopt AI coding assistants trust the assistants’ judgment. 

Beyond coding, Q Developer can assist a business in administering its AWS cloud infrastructure or, at the very least, provide it with the information it requires to do so independently. 

Q Developer can respond to queries such as “List all of my Lambda functions” and “List my resources that are located in different AWS regions.” At this time in preview, the program can also generate AWS Command Line Interface commands (but not execute them) and respond to AWS cost-related inquiries, including “What were the three most expensive services in the first quarter?”

What is the cost of these generative AI conveniences, then?

With certain restrictions, Q Developer is accessible for free in the AWS Console, Slack, and IDEs, including Visual Studio Code, GitLab Duo, and JetBrains. Custom libraries, packages, and APIs cannot be fine-tuned in the free version, which also automatically enrolls users in a data collection scheme. Additionally, monthly restrictions are in place, including a maximum of 25 queries regarding AWS account resources and 5 Agent assignments (such as implementing a feature) per month. (Amazon’s decision to limit the number of queries that can be asked about its services is perplexing, yet here we are.) 

Q Developer Pro, the premium edition of Q Developer, enables users to access additional features such as single sign-on, tools for managing users and policies, higher utilization limits, and IP indemnification. It is priced at $19 per month per user.

The training of the models that support code-generating services like Q Developer frequently occurs on copyrighted or restricted license codes. Not all parties concur with the vendors’ assertion that fair use safeguards them if the models were intentionally or unintentionally developed using copyrighted code. A class action lawsuit accuses GitHub and OpenAI of copyright infringement by permitting Copilot to regurgitate licensed code fragments without attributing the source. 

Amazon states that it will defend Q Developer Pro customers against claims that the service violates the intellectual property rights of a third party, provided that the customers allow AWS to manage their defense and reach a settlement “as AWS deems appropriate.”