Prompt Mode · members-only beta
Type what you want.
ClipMe clips it.
Paste a Kick, Twitch, or YouTube VOD and describe the moment in plain words. ClipMe finds every match across video, audio, and chat, reframes it to 9:16, and captions it — ready to post.
Prompt Mode ships to founding members first. Hands-free clipping is live today.
every time chat explodes
Prompt ModeWhat should you use a prompt for?
Hands-free mode already catches the obvious bangers. A prompt is for the specific thing you already know you want.
Kick IRL streamer
“every time chat explodes”
Chat-velocity spikes, cut and ranked — the moments your audience already voted on.
Gaming streamer
“all my clutch rounds”
Every clutch, captioned and reframed, without scrubbing a 6-hour VOD.
Just Chatting streamer
“my hottest takes tonight”
The rants and reactions that start comment wars — no gameplay needed.
Podcast / react channel
“when the debate gets heated”
Every back-and-forth flashpoint, ready for Shorts.
Clips-channel operator
“the funniest moments of the stream”
A ranked shortlist from someone else's broadcast, posted on your cadence.
Esports team
“every ace and match point”
Team-account content from scrims and matchdays, same day.
VTuber
“every time I broke character”
The breaks and bits your community clips by hand today.
Editor / agency
“all sponsor mentions”
Every brand read pulled for the recap deck in one pass.
How Prompt Mode works
ClipMe watches the whole stream
The engine already scores every second of your VOD across 18 signals — video, audio, and chat together. That full read is what your prompt searches against.
You type what you want
Plain words, no timestamps: “every clutch”, “when chat spams W”, “all sponsor mentions”. One sentence is the whole brief.
Matching moments come back ready
Every match comes back cut, captioned, and ranked — post the best ones as-is. No timestamps, no timeline.
Every match comes back post-ready
A prompt doesn't hand you timestamps to go edit — it hands you finished clips. Face-tracked reframes to 9:16, 1:1, 16:9, word-level captions in 5 languages, and a virality score on every cut so the best one is on top.
- Ranked, not dumped — best match first
- Captions burn in word-by-word
- Auto-post on your cadence, or export
thisalmostwentdown
Two ways to clip
Same engine underneath. Hands-free is the default; a prompt is the steering wheel.
Hands-Free Mode
Live todayDrop a VOD — or connect your channel and let it clip while you're live. ClipMe scores every second and cuts the best non-overlapping moments on its own. Up to ~100 captioned clips per stream, ranked.
Best for: daily volume without touching a timeline.
Prompt Mode
BetaTell it exactly what to hunt for. Because ClipMe reads chat alongside video and audio, prompts about your audience — “when chat loses it” — actually resolve.
Best for: specific asks — a theme, a bit, a sponsor recap.
Apply for the betaPrompt Mode vs Hands-Free
| Hands-Free | Prompt Mode | |
|---|---|---|
| What finds the moment | The 18-signal engine decides | Your words steer the same engine |
| Input needed from you | None — drop a VOD or go live | One plain-English sentence |
| Best for | Daily volume: ~100 ranked clips per stream | Specific asks: a sponsor recap, one running bit, a highlight theme |
| Output | Reframed, captioned, ranked clips | Reframed, captioned, ranked clips |
| Works on IRL & Just Chatting | Yes — chat is a first-class signal | Yes — chat is a first-class signal |
| Availability | Live today, free founding tier | Rolling out to founding beta members |
The engine underneath
Prompt Mode isn't a separate model bolted on — these are the numbers it inherits on day one.
captioned clips from one stream, ranked by virality score
measured benchmark: one 10-hour stream into ~50 ranked clips*
signals scored every second — video, audio, and chat fused into one score
word-level captions, 9:16 / 1:1 / 16:9 reframes
*Measured on 2–4× L40S GPUs; real-world speed varies with stream length, queue and plan.
FAQ
What is Prompt Mode?
Prompt Mode lets you describe the moments you want in plain words — “every clutch round”, “when chat explodes”, “all sponsor mentions” — and ClipMe finds every match in your VOD, cuts it, reframes it to 9:16, and captions it. It runs on the same engine that powers ClipMe's hands-free clipping. Prompt Mode is rolling out to founding beta members; hands-free clipping is what's live today.
What kind of prompts work?
Anything you could tell a human editor in one sentence: reactions (“when I scream”), chat behavior (“every time chat spams W”), topics (“all the moments we talk about the new update”), or business asks (“every sponsor mention”). Because ClipMe reads chat as well as video and audio, prompts about your audience's reaction work too — that's the part upload-first tools can't see. Prompt Mode is in members-only beta.
Which platforms does it work on?
Kick, Twitch, and YouTube VODs — ClipMe is built native to Kick. Hands-free live clipping works while you stream; Prompt Mode — rolling out in the members-only beta — runs against any finished VOD on your account.
How is this different from OpusClip's ClipAnything?
OpusClip's ClipAnything is built for uploaded, finished videos and reads video and audio. ClipMe is stream-native: it also reads your chat, so prompts like “when chat loses it” actually resolve — and it works on Kick, which upload-first tools treat as an afterthought. OpusClip is the better pick for podcasts and finished uploads; ClipMe is built for streamers.
When can I use Prompt Mode?
It's rolling out to founding beta members first. ClipMe is invite-only — Miami creators first, 305 founding memberships. Apply on the beta page or enter a member's invite code to skip the line.
Do I have to write prompts to get clips?
No. Hands-free mode is the default and ships today: drop a VOD or connect your channel and ClipMe cuts and ranks the best moments on its own. Prompt Mode is the steering wheel for when you want something specific.
Describe the moment. Post it.
Prompt Mode rolls out to founding members first — free the entire beta month, founder pricing locked after.