Intelligent Dependence for AI: What to depend on, under which conditions, and for how long

AI is now central to public sector service delivery - but the real question isn’t whether to use it, it’s how to use it wisely. Total “AI independence” is an expensive illusion that risks cutting organisations off from global innovation. The answer is intelligent dependence: making informed, strategic decisions about which AI capabilities to rely on, under what conditions, and for how long. This talk gives public sector leaders a clear, actionable framework for understanding AI dependencies across four layers: infrastructure (compute, data pipelines, resilience), systems & services (models, platforms, adaptability), people & capacity (skills, leadership, knowledge control), and market dynamics (supplier concentration, innovation incentives). Attendees will learn how to map these dependencies, assess risks and opportunities, and avoid drifting into vendor lock-in. Jonny will unpack the AI sovereignty spectrum - from Digital Public Goods and sovereign AI tools to strategic partnerships and targeted open source adoption - and show how to balance immediate capability with long-term adaptability. Real-world public sector case studies will demonstrate how intelligent dependence can protect mission-critical services, reduce costs, and keep future options open. Why this session matters: With procurement cycles closing and AI policy moving fast, leaders face decisions now that will lock in dependencies for a decade or more. This talk arms them with the tools, foresight, and confidence to operate by design, not by default.