AI-native product operating model — Product School talk

What the video covers

Carlos Gonzalez de Villaumbrosia, CEO at Product School, delivers a ProductCon London keynote on what it actually takes to move from AI experimentation to an AI-native operating model. The talk draws on Product School’s own transformation — where 66 percent of their SaaS stack was replaced or removed — to argue that boards are no longer satisfied with pilot projects and now demand measurable return on AI investments.

Who it’s for

Product leaders and senior PMs at organizations stuck in the “messy middle” of AI adoption: past the excitement of initial pilots but before seeing consistent returns. Particularly relevant for anyone building a case for structural team changes or rethinking their SaaS tooling around AI capabilities.

Key takeaways

1. The productivity J-curve is real. Moving from AI pilots to production environments often causes a temporary dip in team output before gains materialize. Gonzalez argues that companies fail not because AI does not work, but because leadership loses patience during this valley. Understanding the J-curve helps PMs set realistic expectations with stakeholders and build a timeline that accounts for the learning period.

2. Roles are converging into “builders.” Traditional boundaries between product managers, designers, and engineers are collapsing. PMs at Meta are informally changing their LinkedIn titles to “AI builder.” The rise of vibe-coding, vibe-research, and vibe-design means a single person can now prototype, research, and ship in ways that previously required a full cross-functional team.

3. The one-pizza team is shrinking further. AI agents “eat tokens, not pizza,” which means teams can operate with fewer people while increasing throughput. This creates pressure on team sizing and hiring models. Gonzalez describes a future where a two-person team augmented by AI agents outperforms traditional five-person squads.

4. Replacing your own tools is part of the strategy. Product School forced an AI-native workflow by cutting two-thirds of their existing SaaS tools. The argument is that incremental adoption does not work — organizations need to remove the comfortable fallback of legacy tools to push teams into building AI-first processes.

5. The AI leadership formula has four dimensions. Gonzalez proposes evaluating product leaders across Producer, Expert, Leader, and Manager quadrants to assess their readiness for AI-native work. The framework helps organizations identify which leaders can drive transformation and which need upskilling.

Worth watching if…

Your organization has completed AI pilots but struggles to scale them into production, and you need a strategic framework to present to leadership about what structural changes are required to move forward.