Multimodal & Media
ANTS Platform handles multimodal AI: when your traces include images, audio, video, or documents, the platform stores that media and renders it inline so you can review exactly what went into and came out of a call.
Media is captured automatically when present in a trace's input, output, or metadata — there's no separate UI page. It renders within the trace detail view.
What's supported
A wide range of content types, including:
- Images — PNG, JPEG, WEBP, GIF
- Audio — MP3, WAV
- Video — MP4, WEBM
- Documents & data — PDF, JSON, TXT, and more
How it works
When you send a trace that contains media, the SDK uploads the asset to secure blob storage and references it by a media ID. In the trace view, the platform resolves that reference to a temporary signed URL and renders the image, audio, video, or document inline. Assets are validated (content type, size, and integrity hash) on upload.
Media is stored in the blob storage configured for your deployment. Signed URLs expire, so media is only accessible to authenticated users with access to the project.
Next steps
- Traces & Observations — where media renders.
- Sending Data In — instrument your app to capture multimodal content.
- Data Privacy — storage and access considerations.