42:I[185,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"SendEmailButton"] 44:I[63625,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"AddToLibraryButton"] 45:I[29721,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"ShareButton"] 46:I[58016,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"ProcessEpisodeButton"] 47:I[45269,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"SignupCTA"] 3a:["$","time",null,{"dateTime":"2026-06-25T13:15:00.000Z","children":"June 25, 2026"}] 3b:["$","span",null,{"aria-hidden":true,"children":"·"}] 3c:["$","span",null,{"children":"00:08:17"}] 43:T477,

Nikolay Donets, Head of Machine Learning Engineering at Revolut, on what it takes to run AI across more than 70 million customers, 200+ products, and 40+ countries - and why the hard part is no longer the model but the control plane around it: one gateway, a use-case-based governance layer, fallback chains, cost controls, and mandatory human oversight. Recorded at RAAIS 2026.

Chapters:

0:00 Intro - Revolut's AI at scale

1:24 The problem: classical ML and three libraries

2:54 The 2022 shift to API-served models

4:25 Four internal groups, four sets of needs

9:39 The decision: govern the use case, not the model

10:54 One central gateway vs. distributed libraries

14:10 Performance monitoring and drift detection

17:33 Lesson: fallback chains and the silently-dead model

20:02 Lesson: frontier vs. non-frontier cost (up to 8x)

20:48 Lesson: the platform is the org chart

22:59 Case study: from Rita to AIR

26:40 Voice support at scale

28:21 AIR, the in-app assistant

30:25 Q&A: human oversight, hallucinations, AI as judge