Apple Siri AI taps Google Gemini to power agentic assistant

By rebuilding Siri on top of a customized Gemini model and exposing it as a first‑class, cross‑platform agent, Apple is effectively conceding that frontier capabilities currently live outside its own models — but it’s also weaponizing its distribution. A Gemini‑powered Siri AI that can see your screen, understand personal context across apps, and execute multi‑step tasks turns 2+ billion Apple devices into front ends for advanced agentic systems.([m.ajupress.com](https://m.ajupress.com/amp/20260609150970214))

Crucially, Apple isn’t locking that power to one model. Between Siri AI and the new Foundation Models framework, users and enterprises can increasingly choose whether Gemini, Claude or ChatGPT sits behind their assistant, while Apple controls the UX, guardrails and on‑device context. That weakens single‑vendor moats and pushes competition toward better agents and workflows rather than raw model size. For the race to AGI, this accelerates the feedback loop: more real‑world multimodal interactions, more complex tool‑calling, and richer behavior data flowing back into model training — even if Apple itself isn’t training the largest frontier models.