SDKv2.2.0

Agents now support media-only inputs

October 23, 20251 min read

Agents can now process inputs that contain only media (no text). This unlocks use cases such as camera-to-answer, voice-only prompts, and file-first interactions, reducing friction in multimodal experiences.

Details

Accepts images, audio, video, or files without requiring text
Simplifies mobile and voice-first integrations

Who this is for: Teams shipping multimodal or mobile-first experiences that prioritize media over text.

Shipped around the same time

Unblock reuse and governance by decoupling knowledge from agents and teams

Establish consistent multi‑agent behavior with an experimental Culture Manager and shared knowledge space

Smarter workflow decisions with session-aware routing and conditions