A five-slide brand carousel from one chat message
The Versely Agent pitches carousel directions, writes every slide caption before a single image exists, then renders the set with the text baked in.
The Versely Agent pitches carousel directions, writes every slide caption before a single image exists, then renders the set with the text baked in.
Drop a product URL into the Versely Agent chat: it reads the listing into a brand kit, re-casts a public workflow template with your product, quotes the render, then writes and schedules the posts.
Pasting a brand URL can draft a campaign plan. It will not replace a human brief with constraints.
Paste a link, get a shot list. The analyzer is research, not a finished film.
Scenes as columns of nodes are a plan graph. An NLE timeline is where frames meet.
The brain plans; the catalog renders. Pin GLM 5.3, Grok 4.6, Kimi, Claude, Gemini — then name Kling or Seedance separately.
Same attachment rule as GLM: no stills on the wire. Cost-efficient when the brief is already words. Not a video model.
Pin z-ai/glm-5.3 behind the Versely agent. Text-only, 1M context, reasoning always on. Name Kling or Seedance in the same brief. Do not attach a still.
Generate a sound effect is one agent job: a described hit, riser, or ambience. Using generate_music for SFX, or regenerating a video for one missing hit, wastes credits.
Auto-captioning is one named job on audible speech. A fixed overlay, a transcript-only file, or captions on a silent clip spend this row on nothing.
add_video_captions stamps one hook, headline, or CTA on the whole file. Transcription is a different job, and overlaying a miss wastes credits.
create_hook_pack is priced as a pack, not per hook. That is why 5 distinct openings beat 5 separate generates.
Viral-post analysis is one named research job at 1 credit. Using it to judge your generate, or using review_generation to explain someone else's Reel, answers the wrong question.
Get feedback on a video is one agent job: frames, beats, on-screen text, optional transcript. Grading a generation against its prompt is a different job.
Check or resume a stuck workflow run is one agent job. get_workflow_run_scenes diagnoses a scene. Cancelling a healthy run to start over wastes completed-scene credits.
Add music or a voiceover to a video is one agent job. attach_audio_to_video lays an existing file. Generating speech or music is a different job.
Save and reuse a video workflow is one agent job. save_workflow keeps scenes and assets. Running a one-off clip is a different job.
Edit my slideshow is one agent job: add, remove, replace, or reorder. Regenerating the carousel, or converting to video first, is a different job.
Voice change is one named audio job: same delivery, different Cartesia voice. TTS from scratch, cloning, and dubbing are other bills.
Stitching is one named edit job: images and clips in a given order. Asking it to invent footage spends a compose fee and still leaves the shots unmade.
create_content_pipeline plans a multi-step task from a brief. A single generate is a different job, and running it as a campaign wastes credits.
create_custom_voice is Qwen voice design plus one read. Cloning a reusable voice_id is a different job, and designing every line wastes credits.
A branded hook pack is one named job: N distinct 1080p portrait opens, one charge. Using it as a movie maker or a single-hook renderer wastes the pack.
create_slideshow plans and fills slides from a prompt. Uploading your own photos is a different job, and mixing them wastes credits.