AI-driven software development holds great promise. But so far, the productivity results have been disappointing. Despite this, engineering managers are being directed to increase AI usage and spending.
How to bridge this gap? We need a way to better steer AI agents and thus avoid the off-course “drift” that requires laborious downstream correction.
Spec-driven AI development can be an answer to this productivity crisis – it shifts the focus from just generating more code faster to capturing clear intent in shared specifications. These specs guide AI to produce consistent, reliable implementations tied to business goals. As a result, collaboration gets easier, and teams are able to deliver more predictable, repeatable results.
We’ll demonstrate a complete, spec-driven development loop inside IntelliJ IDEA with JetBrains AI. Starting from a requirements doc, tickets, and meeting notes, we’ll use AI to draft an implementation plan, generate the initial code, and create related tests automatically. Then, we’ll refactor and validate the code against the original spec.
Trisha Gee is a Java Champion, author, and internationally recognized speaker with over two decades of experience in software development. Known for her deep expertise in Java, high-performance systems, and developer productivity, Trisha has worked as a developer and leader in organizations ranging from startups to global enterprises. She's passionate about sharing knowledge and helping developers write more expressive and efficient code.
Trisha is the author of multiple technical books, including Head First Java (3rd Edition) and Getting to Know IntelliJ IDEA, and she frequently contributes to developer communities through blogs, webinars, and international conferences. She's also a strong advocate for grassroots learning and regularly supports local user groups and meetups to help developers connect, grow, and thrive.
When she's not writing or coding, she's championing inclusive practices and mentoring the next generation of developers.