AI tools promise dramatic shortcuts in coding, testing, and delivery. The temptation is strong: drop the "slow" or difficult practices of Extreme Programming and let the models do the heavy lifting. This session argues the opposite. Precisely because AI accelerates so many steps, the core XP practices become more valuable, not less. Without them, teams risk losing the tight feedback loops, shared understanding, and continuous validation that keep quality high and products aligned with real needs.
Drawing on real-world experience in an organization that both ships impactful software products and builds capability through a "teaching-hospital" model for software-development teams, we’ll examine four practices through an AI lens:

Pairing: How collaboration prevent the silent accumulation of opaque, low-quality code
Test-Driven Development: How TDD keeps code clean, enforces YAGNI, and remains the most reliable way to keep AI-generated code honest
Cross-functional/whole-team approach: The necessity of keeping product, design, and engineering in the same feedback loop when AI can generate large volumes of work in isolation
Small incremental releases and rapid product validation: How rapid AI output makes frequent, validated slices of functionality even more critical to avoid building the wrong thing at high speed

Attendees will leave with a clear rationale for applying (and adapting) XP practices in an AI-augmented world, plus practical ways to integrate AI tools without sacrificing the quality and learning loops that XP enables.