What is an AI product manager — Mind the Product talk

What the video covers

Five members of the One NZ (formerly Vodafone NZ) AI product team present at ProductTank Auckland. They describe one of New Zealand’s largest AI transformation programmes — spanning finance, customer service, engineering, and back-office — with almost 50 new AI solutions launched in 2025. The speakers include the GM of AI Product, the AI Product Chapter Lead, and three AI PMs who each present specific case studies from their work.

Who it’s for

PMs who have just moved into an AI product role or are considering the transition. The talk provides ground-level perspective from practitioners who are learning as they go, rather than polished frameworks from consultants. Particularly useful for PMs at telcos, utilities, or other large service companies running broad AI transformation programmes.

Key takeaways

1. AI product management is about shaping the space around the product, not defining the product itself. Steph, the PM behind One NZ’s AI concierge for over one million mass-market customers, describes how her role shifted from managing features to managing ambiguity, trust, and failure modes in real time. What started as a simple question — “can we answer customer queries and deflect calls?” — turned into an evolving system that requires constant shaping and constraint.

2. You are never done. Unlike traditional product development where features ship and stabilize, AI products continuously change as models improve, data shifts, and edge cases surface. The PM’s job is not to reach a finished state but to build processes for ongoing monitoring, evaluation, and intervention. The team learned this through experience with their AI concierge, where unexpected failure modes appeared long after launch.

3. AI moves faster than your ability to define it. Capabilities change so rapidly that rigid roadmaps become obsolete within weeks. The team adopted a mindset of asking “where will this fail?” instead of “what should this do?” — a fundamental shift from feature-specification thinking to failure-mode thinking.

4. The “AI product manager” is just a product manager who works with AI. The team pushes back against treating AI PM as a separate discipline. The core skills — user empathy, prioritization, stakeholder management — remain the same. What changes is the technical context: PMs need enough understanding of models, data, and evaluation to have productive conversations with their engineering counterparts.

5. Blending product and engineering craft is the emerging model. The AI Product Chapter Lead describes building a “workbench” for AI product management that combines traditional product tools with engineering experimentation. The line between product definition and technical implementation is deliberately blurred.

Worth watching if…

You are transitioning into an AI PM role and want to hear from practitioners who describe both what went right and what went wrong, rather than from those who only present success stories.