In last year's "First Agile, Then Agentic" talk, I argued that organisations need the right foundations before adopting agentic development and that you merely create a new bottleneck if all you do is speed up implementation work.
Studies show that AI makes coding faster. But the data suggests that for organisations without the right foundations, delivery doesn't get faster. Overall cycle time increases and software stability decreases. Yet many organisations respond by reinvesting in more output: accelerating requirements engineering and shipping more features. In this talk I will explain why that’s the wrong move.
The real opportunity is product discovery: understanding what is worth building in the first place. I will take a critical look at current AI discovery tools, examine where they hit fundamental limits, and make the case for investing the freed capacity in deeper, more rigorous discovery work. I will also explore what this means for how teams are structured and why agentic development finally makes it possible for developers to contribute where their technical perspective creates the most value.
