Google shifts to Interactions API for Gemini models and agents
Google DeepMind has quietly upgraded the way developers connect with its Gemini models and agents. Starting this week, the Interactions API is now the default interface, replacing the older generateContent API. The move simplifies how applications send and receive data by using a typed-step schema instead of the previous role-based structure. New agent features will ship exclusively through this updated API going forward.
A cleaner contract for AI services
The change signals Google’s push toward a more consistent and maintainable way to handle multi-turn conversations. The Interactions API describes each exchange as a sequence of typed steps—such as user messages, model responses, and tool calls—rather than assigning roles like “user” or “assistant.” Developers benefit from stricter schema validation and clearer error messages, which should reduce integration headaches.
What developers need to know
Existing apps can keep running on the legacy generateContent API for now, but Google advises migrating sooner rather than later. Code samples and documentation have been updated to showcase the new pattern, and the company says the Interactions API is already powering the latest agent capabilities. Teams that adopt the change early will gain access to upcoming features without retrofitting later.
For organizations building with Gemini, the transition is less about major rewrites and more about adopting a cleaner contract. Google’s move aligns with broader industry trends toward structured, typed interfaces that scale more predictably across different use cases.
Source: The Decoder. AI-assisted editorial synthesis — TechnoExpress.