Let’s take a moment to remember CodeWhisperer, Amazon’s AI-powered assistive coding tool. As of today, CodeWhisperer is no longer functioning—sort of.
Q Developer, a member of Amazon’s Q family of business-oriented generative AI chatbots, is now CodeWhisperer. Q Developer is also part of the newly announced Q Business. Accessible via AWS, Q Developer assists developers with various tasks they encounter in their daily work, such as debugging and upgrading applications, troubleshooting, and conducting security scans. It offers similar functionality to CodeWhisperer.
In an interview with Eltrys, Doug Seven, GM and director of AI developer experiences at AWS, suggested that CodeWhisperer didn’t quite hit the mark in terms of branding. Third-party metrics indicate that CodeWhisperer faced challenges in keeping up with its main competitor, GitHub Copilot. GitHub Copilot boasts over 1.8 million paying individual users and a significant number of corporate customers, while CodeWhisperer struggled to gain the same level of momentum despite offering a free tier. (The initial impressions were definitely not favorable.)
“CodeWhisperer is where we began with code generation, but we desired a brand and name that would be suitable for a broader range of applications,” Seven explained. “Q Developer is the next step in the progression of CodeWhisperer, expanding its scope to encompass a wider range of capabilities.”
In order to achieve this, Q Developer has the capability to generate code that includes SQL, a widely used programming language for database creation and management. It also provides tools for code testing, as well as support for the transformation and implementation of new code derived from developer queries.
Customers can customize Q Developer to improve the accuracy of its programming suggestions for their internal codebases, just like a database administrator. This option was also provided by the now-deprecated CodeWhisperer. With the help of agents, Q Developer has the ability to independently carry out tasks such as implementing features and restructuring code, as well as documenting and refactoring.
If you make a request to Q Developer to add a ‘add to favorites’ button in your app, they will carefully analyze the app code, generate new code if needed, create a detailed plan, and thoroughly test the code before making any changes. Before implementation, developers can thoroughly review and make adjustments to the plan. They can seamlessly connect the necessary steps and incorporate updates throughout the relevant files, code blocks, and test suites.
“What occurs in the background is that Q Developer creates a development environment for code development,” Seven explained. When it comes to feature development, Q Developer manages the code repository by creating a branch, analyzing it, and making the necessary code changes as requested before returning them to the developer.
Agents have the capability to automate and oversee code upgrading processes, as stated by Amazon. Java conversions, specifically Java 8 and 11 built using Apache Maven to Java version 17, are currently available. .NET conversions will be available in the near future. “Q The developer carefully examines the code, searching for any areas that require upgrading.” Seven explained that after identifying the necessary changes, they return the code to the developer for their review and final approval.
Agents bears a striking resemblance to GitHub’s Copilot Workspace, in my opinion, as it also generates and executes plans for bug fixes and new features in software. And, similar to a database administrator, I remain skeptical that this independent approach can effectively address the challenges associated with AI-powered coding assistants.
GitClear’s analysis of over 150 million lines of code committed to project repos in recent years revealed that Copilot was causing an increase in the push of erroneous code to codebases. Additionally, security experts have cautioned about the potential for Copilot and similar tools to exacerbate preexisting bugs and security vulnerabilities in software projects.
It’s not unexpected. AI-powered coding assistants are quite impressive. However, their training relies on existing code, and patterns from other programmers’ work influence their suggestions, sometimes leading to serious flaws. Guesses made by assistants can introduce bugs that are challenging to identify, particularly when developers heavily rely on the judgment of AI coding assistants.
With expertise in managing cloud infrastructure on AWS, Q Developer can assist companies in effectively handling their systems or provide them with the necessary information to manage them independently.
A developer can handle requests such as “Provide a list of all my Lambda functions” and “Retrieve a list of my resources located in different AWS regions.” Currently in preview, the bot has the capability to generate AWS Command Line Interface commands and provide answers to AWS cost-related queries, such as identifying the top three most expensive services in Q1. However, it does not have the ability to execute these commands.
Could you please provide information on the pricing for these generative AI conveniences?
Various platforms, such as the AWS Console, Slack, and IDEs like Visual Studio Code, GitLab Duo, and JetBrains, offer free access to Q Developer. However, it is important to note that there are certain limitations. The free version does not offer the ability to customize libraries, packages, and APIs, and it automatically includes users in a data collection program. Monthly limits are in place, which include a maximum of five tasks for agents per month (such as implementing a feature) and 25 queries regarding AWS account resources per month. (I find it puzzling that Amazon would set a limit on the number of questions one can ask about its own services, but this is the reality we face.)
Q Developer Pro, the premium version of Q Developer, is available for $19 per month per user. This upgraded version offers higher usage limits, user and policy management tools, single sign-on functionality, and most notably, IP indemnity.
Often, restrictive licenses or copyrighted code train the models supporting code-generating services like Q Developer. Some argue that the development of a model using copyrighted code, whether knowingly or unknowingly, may protect vendors through fair use. However, there are differing opinions on this matter. GitHub and OpenAI are suing in a class action motion alleging copyright infringement. The lawsuit claims that Copilot, their software, reproduces licensed code snippets without proper attribution.
Amazon has stated that it will provide support to Q Developer Pro customers in the event of claims related to infringement of intellectual property rights. However, it is important to note that AWS will have control over the defense and settlement process, making decisions as they see fit.
