A year ago, AI was still moving faster than everyone else, outpacing companies, governments, and just about anyone trying to keep up. By the end of 2025, that finally started to change. In this episode, our host Leslie D'Monte and Shouvik Das recap 2025 - the year AI stopped being just a breakthrough story and became a governance story. As generative models gave way to agentic systems, and as frameworks like India’s DPDP Act came into sharper focus, AI entered a new phase, one where rules, guardrails, and accountability mattered as much as capability. The conversation looks at how 2025 forced a reset. Enterprises rushed to deploy GenAI and agents, only to encounter hallucinations, broken integrations, and the limitations of legacy systems. At the same time, the question shifted from what AI can do to where it should actually be used. A big part of the episode zooms out to India’s approach. Instead of heavy regulation, the focus has been on setting boundaries about content labeling, competition oversight, and applying existing laws while still pushing investments in GPUs, data centres, and Indic language models. The goal isn’t dominance; it’s relevance. And then there’s the bigger test: can AI work outside the bubble? In a country where most people don’t live in metros or use cutting-edge devices, success looks less like chatbots and more like IVRs, public services, healthcare tools, and systems that disappear into daily life. By the end, the takeaway is simple: 2025 wasn’t about AI replacing humans or reaching artificial general intelligence. It was about learning where AI fits, where it fails, and how human judgment, guardrails, and local context still matter more than raw capability. Learn more about your ad choices. Visit megaphone.fm/adchoices