Breakfast – AI Kill switch

Breakfast, meet control.
Coffee, pastries, and a morning of straight-talk engineering on shipping fast and the AI Kill switch.

When: 6 October 2026, 8:30–10:30 CET
Where: Amsterdam, TBD
Host: Wojtek Gawroński — Developer Relations, Unleash

On the Menu:

• Why now? The market moved from fear of messing up to fear of missing out. AI isn’t just a feature – it’s an amplifier for how software gets built, but leadership and engineering carry the risk.
• Kill switch for AI is a necessity. You need a way to turn off misbehaving AI, or the code it wrote, for a smaller blast radius, back to a known safe state, with no disruption to surrounding systems and no unnecessary delay.
• But that’s just the start. In regulated domains, it was never about disabling a feature in anger. Can you recover in seconds? Document why and how? Target precisely enough to prevent it? Plus the daily reality: capacity, load shedding without sacrificing VIPs, cost and quality experiments, AI-powered results customers can trust.
• FeatureOps is the structured answer. Build runtime control and progressive delivery from the ground up, adopting it as a strategic organizational practice rather than a tactical on/off switch.
• Join the case studies review and open discussion. Lloyds Bank and Visa are shipping change safely at bank and card-network scale without a rewrite. Dive deeper in Q&A with the Unleash team.

Topics:

• AI kill switches & governance — How to give every AI-driven feature a runtime off-switch, and prove to your governance team that it works before an incident forces the question.
• Case studies: Lloyds Bank & Visa — Real production stories on rolling out change safely at bank- and card-network scale, without a rewrite.
• Open floor — Q&A and coffee refills with the Unleash team and fellow engineers running production in Amsterdam.