You built AI agents. How do you keep them under control? Ani Narayan, co-founder and CGO of Lyzr AI

What happens when AI agents spread across your business, each using different models, tools and data?I spoke with Anirudh (Ani) Narayan, co-founder and CGO of Lyzr AI, about the work involved in taking agents into everyday operations. We discuss where businesses are using them, how teams prioritise ideas and why some workflows need custom integration around legacy systems.Ani describes Lyzr’s approach to agent infrastructure and observability, including Opencontroller, the control plane for AI agents. We also explore an insurance company case study that keeps a human at the front of a sensitive customer journey while using AI behind the scenes.That leads into a practical discussion about when existing tools are enough and when a business needs something purpose-built.Guest: Anirudh (Ani) Narayan, co-founder and CGO, Lyzr AIHost: Kane Simms, VUX AICHAPTERS00:00 Welcome00:46 From an agent builder to enterprise infrastructure02:42 The agent-management challenge04:36 Where businesses are deploying agents07:14 Cost, flexibility and privacy12:02 Consulting, marketing and operational use cases17:19 Educating teams and prioritising ideas19:16 Keeping humans in sensitive customer journeys20:45 Kane’s AI video-editing experiment24:20 When custom agents make sense26:54 Observability and the control plane31:41 What AI providers and customers are getting right37:21 Automating the podcast workflowLINKSLyzr AI: https://www.lyzr.ai/Opencontroller: https://www.lyzr.ai/opencontrollerAni on LinkedIn: / anirudh-narayan Customer examples, figures and product observations reflect the speakers’ accounts at the time of recording.Subscribe for conversations about AI, customer experience and the practical work of deploying agents.Follow Kane on LinkedIn: / kanesimms Find out more about VUX: https://vux.aiSubscribe to The AI Ultimatum Substack: https://open.substack.com/pub/kanesimms