As generative AI continues to develop, experts expect that emerging technology to support all aspects of the software development lifecycle, from backend development to testing and maintenance tasks.
On a basic level, developers can currently “use generative AI platforms to generate small snippets of business logic for the APIs that they’re developing,” said Aju Mathew, Vice President of Software Engineering at Aspire Systems. But looking ahead, developers will begin to use it in a more advanced way—for example, deploying AI to generate all the backend APIs requested in the design document. This upper-level work would require advanced prompt chaining and parameterization, which suggests that AI will be performing tasks requiring a high level of development skill.
Experts also believe that AI’s transformation of the software development process will happen sooner than some might think. Listen to my extended interview with Aju Mathew to learn about the myriad AI techniques and tools—some of them already emerging—that software developers are likely to adopt.
Watch the full interview with Aju Mathew, or jump to select highlights below.
In the language of software development, greenfield development refers to building a brand new application. In contrast, brownfield development refers to revising code that supports an established system.
Brownfield is, arguably, the more challenging of the two, since the new code must interoperate with the idiosyncrasies of a legacy infrastructure. Yet generative AI can support even some forms of brownfield development.
“Developers can use generative AI platforms to generate fixes for the code issues or basically [create] incremental business logic for any API’s functional changes,” Mathew said. Expect these capabilities to advance with time.
Tools like Amazon Q, which can reverse engineer code, support this functionality already. These tools can “extract the business logic and document it, and this is useful for understanding legacy code bases,” he said.
To speed up workflow, users can automate the unit test cases for the backend APIs, which get generated using the specified language framework. “For example, if it’s J unit for Java and N unit for .NET, we can generate API or swagger documentation,” Mathew said. For software migration work, developers will use generative AI to upgrade complete software technology frameworks.
“Going forward, I would anticipate developers being able to do much more with generative AI platforms and software development, probably even generate complex business logic,” Mathew said. Advances like these are “futuristic, but possible.”
Read our guide to generative AI vs. AI to learn about the advantages, limitations, and ethical considerations of working with these dynamic technologies.
The post Aspire’s Aju Mathew on AI and Advanced Application Development appeared first on eWEEK.
