Beyond code generation: Five lessons on AI, reliability, and recovery
At a TechSparks 2026 roundtable hosted by OPPEX, technology leaders discussed the challenges of maintaining AI-generated code, managing risk, and responding when production systems fail.
The 2026 TechSparks roundtable, moderated by Madanmohan Rao of YourStory, brought together technology leaders to explore AI's impact on software development and the challenges it presents. Abhishek Dwivedi, VP Engineering at Cimpress, highlighted that while AI speeds up code generation, it doesn't automatically improve maintainability or troubleshooting.
Amit Khanna, Global Head of Engineering at Amadeus, pointed out that faster code isn't necessarily better if it leads to more business issues. Anand Rangarajan, VP Engineering at Google DeepMind, emphasized that AI agents should have varying levels of autonomy based on the risk involved. Sridhar Jonnala, Chief AI Scientist at IBM, stressed the importance of context for AI agents to function effectively, suggesting they need access to knowledge graphs and shared repositories containing information about system interactions and past incidents.
The roundtable concluded by stressing the need for human oversight, particularly in regulated industries, and the necessity of establishing feedback loops to improve AI-generated code and incident response processes.
Written by urgent.news from YourStory's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
