GIDS news media, articles, insights and virtual events educate and illuminate its audiences so they can be fully prepared to deal with the new realities at work and in their professions.
Get the latest event updates, and insights from today's leading voices.
< session />
Thu, December 10AI-Native Software DevelopmentProduction AI Systems
As organizations explore the idea of software factories, there is growing interest in using AI to industrialize software delivery. However, coding agents introduce a fundamental challenge: unbounded intelligence creates an unbounded surface to verify. While that trade-off may be acceptable for everyday development, it becomes problematic when the goal is to produce repeatable outcomes for a specific class of problem.
This session argues that industrialized software delivery requires purpose-built harnesses that bound agent behaviour. The agent explores potential solutions while the harness adjudicates the results. Using a performance profiling and optimization harness as the primary example, we will examine a system in which an agent identifies hotspots and proposes fixes while the harness determines whether the improvement is real. Each run either produces a measured gain or leaves the codebase unchanged. Regressions are discarded and retried a bounded number of times before the process stops.
The talk is structured around three design principles. First, bounded intelligence through the separation of exploration and adjudication. Second, invariants and correctness, focusing on what must remain true after an agent acts. Third, trust as a design constraint, with an emphasis on making evidence inspectable until confidence in the system is earned.
Attendees will leave with a practical understanding of how purpose-built harnesses can be designed to support repeatable outcomes and how the underlying principles can be applied to their own software delivery workflows.
What You Will Learn:
Who Should Attend:
< speaker_info />
Ragunath Jawahar is the Founder of Legacy Code HQ, where he specializes in helping developers and organizations master massive, complex codebases. With nearly 15 years in the industry and 5 years working with large codebases across startups and enterprises, he discovered that software complexity is fundamentally a human comprehension problem, not just a technical one.
To solve this challenge, Ragunath has built innovative visualization tools including Eureka and Timelapse (open-sourced on GitHub), which help developers navigate complex systems by surfacing relevant information while filtering out noise. His unique expertise combines legacy codebase rescue with 2+ years of AI-assisted development experience, positioning him to address a critical emerging problem: AI's acceleration of generating hard-to-maintain codebases.
Through his work at Legacy Code HQ, Ragunath teaches developers how to harness generative AI to build production-grade applications while avoiding maintainability pitfalls—leveraging first principles from human cognition, software development, and AI. This rare combination of legacy code mastery and AI expertise makes him uniquely qualified to help teams build maintainable software in the age of AI acceleration.
