Kartik Hosanagar

Why isn’t AI paying for itself yet? Many rollouts hand people faster tools and stop there, so the work still waits in the same approval chains. Hosanagar takes audiences through four stages: people get faster, the work is redesigned, AI takes on whole tasks by itself, and teams and decisions are rebuilt around it. In his example, one piece of marketing content took three months to pass through an analyst, a specialist, the creative team, brand review, and legal. Give everyone in that chain AI, and the piece still waits in every queue. Redesign the work so AI drafts and a marketer decides, and individual speed finally becomes the organization’s speed. Attendees leave knowing which stage their organization is in and what to change first.

If someone asked AI which organization to choose in your field, would it name yours? People are starting to send AI to research, compare, and sometimes buy. When he asked one keynote audience who saw AI as a new channel, every hand rose. “It’s your new customer,” he told them, one with its own way of deciding. “You have spent decades figuring out how to inform and persuade your human customers but you need to now learn how to inform and persuade your AI customer,” he says. He shows audiences how to see what AI says about them today, find the sources shaping its answers, become one of those sources, and understand how it reasons. The same shift reaches sales teams whose customers research with AI before any call, and organizations whose members, patients, students, or donors now ask AI first. Attendees leave knowing how to earn the recommendation.

Where is the edge when every competitor has the same AI? Hosanagar compares AI to electricity: once every organization had it, the edge was what each did with it. Every rival can rent the same AI models, so “building on top of a shared model is not a strategy,” he says. The advantage comes from what only one organization holds: its own data, its own know-how, and work built to use them. Any insurer can send a claim to AI, he notes. The winner knows which claims to automate and which still need a person. Attendees leave able to name their own lasting advantage and to decide what to build first.

What happens to your job and your skills as AI does more? This is Hosanagar’s talk for every seat in the room. “The effort we avoid is often the expertise we lose,” he writes. Leaning on automation can quietly wear down a skill, as evidence from cockpits, clinics, and classrooms shows. In one of his experiments, people who kept a creative role did better work than people who only approved AI’s drafts. He shows how to keep that role: do the first pass yourself, build in review points, and practice without AI. Attendees leave knowing how to stay sharp and stay in charge of their careers.

Who is accountable when AI makes the decision? This is Hosanagar’s talk on responsible AI and governance. As AI agents take on real tasks, someone has to decide what they may do alone, how to check their work, and who answers for the result. He has argued for keeping people in control since his 2019 book proposed an algorithmic bill of rights. In a 2026 study he co-wrote, top AI models gave complete answers to business-case questions only about half the time. That is why a person still has to check the work. Attendees leave ready to set limits on AI in their own organization.