Media & AI

Generate and analyze images, audio, and video. Most of these call an AI model, so you pick a model when you run them (on the Test tab) — and they use credits. video_extract_frames is the exception: it processes video locally (FFmpeg), no model or credits.

generate_image

Generate an image from a text prompt (DALL·E 3/2, Imagen 4.0). Category: Action · Setup: a model.

• Inputs: prompt (required), size, style (natural/vivid), quality (standard/hd), model.

vision_analyze

Analyze an image — OCR, charts, receipts, general Q&A. Category: Data Pull · Setup: a model + an image.

• Inputs: an image via image_url / image_base64 / file_id / artifact_id, plus prompt, model.

text_to_speech

Convert text to natural speech audio (multiple voices/languages). Category: Action · Setup: a model.

• Inputs: text (required), voice, language, speed, format (mp3/wav/…), model.

audio_transcribe

Transcribe audio to text (Whisper, Deepgram, Google, Azure). Category: Data Pull · Setup: a model + audio.

• Inputs: audio via audio_url / audio_base64 / attachment_id / file_id, plus language, output_format, task.

video_extract_frames

Extract frames from a video for analysis. Category: Data Pull · Setup: a video (no model). Pairs with vision_analyze.

• Inputs: video_url or attachment_id, fps (capped at 20 frames), include_thumbnail, output_folder_id.