Skip to main content
R
RelatiV[Φ]
v2.0 · live now·Φ = taste, not length

Turn one video into
ten viral clips
in 60.0s.

RelatiV watches, listens, and finds the moments that hook — then writes the caption, the title, the hashtags.You paste a URL. You publish clips.

See the engine
97.3%
caption accuracy
4.1 min
for 2h video
10 clips
per source
8.4×
faster than manual
STEP 01 · 02 · 03 · 04

Paste a URL.
Get ten clips.

No upload, no editor, no waiting for an export queue. Drop a YouTube link, watch the engine work, download ready-to-publish clips.

1

Transcribe

Whisper · RunPod

2

Energy peak detect

librosa · 12 Hz

3

Hook score

Claude · emotional peak

4

Render

YOLO · 9:16 frame

Your clips will appear here

Paste a YouTube URL or load a demo

THE ENGINE · 4 STAGES

Four models, one decisive verdict.

Each stage is replaceable. The pipeline is not. Run any video through it and the same ten moments surface.

Φ-Speech

Transcribe, serverless

4 min
for a 2-hour video
// faster-whisper · RunPod
const transcript = await transcribe(video, {
  model: "large-v3",
  language: "auto",
  word_timestamps: true
});
// → 1.2M words · 4,820 turns
math:log P(wₜ | w<t) → softmax

Φ-Hook

Claude picks the 10 winners

97%
caption accuracy
// Claude Haiku 4.5
const scores = await score(moments, {
  rubric: ["emotional_peak", "narrative_complete", "shareability"],
  top_k: 10
});
// → 4 of 4 judges agree
math:argmax Φ(x) · Φ ∈ ℝ⁺

Φ-Frame

YOLO re-frames for 9:16

<2px
centering error
// YOLO v8 active-speaker
const face_box = await detect_active(face, {
  smoothing: "kalman",
  aspect: "9:16",
  margin: 0.08
});
// → 47.2 fps · 12 ms latency
math:x' = σ(Wx + b)

Φ-Story

Captions, titles, hashtags

faster than manual
// Anthropic + word-by-word
const clip = await render(moment, {
  captions: "word-by-word",
  titles: "viral-A/B/C/D",
  hashtags: "trending + niche",
});
// → 4 hook variants
math:L = -Σ yᵢ log ŷᵢ
HOOK DETECTION · LIVE

We don't guess.
We measure.

Every 250 ms, the engine scores the moment against 4 axes: emotional peak, narrative completeness, shareability, attention spike. Only the top 10 cross the Φ = 0.70 threshold.

Emotional peak
0.42
Narrative
0.31
Shareability
0.18
Attention Δ
0.09

energy(t) · 2-hour source

audio_rms + speech_dynamics

detected peaks

3 / 7200s

Φ-threshold = 0.70Φ=0.91Φ=0.96Φ=0.930:001:32:00
signal:E(t) = ∫ audio_rms · w(τ) dτ
USE CASES

Built for people who ship content.

01

Podcasters

12 min

Auto-detect the moment your guest says something quotable. Get 10 ready-to-post clips before the episode ends.

02

Sports Highlights

8 sec

Match events synced to commentator energy spikes. Highlight package ready seconds after the whistle.

03

Coaches & Educators

4.7×

Long lessons, short attention spans. Find the 60-second insight that drives saves and shares.

04

Live E-commerce

+18% CVR

Pin the moment of the product reveal, the demo, the discount drop. Conversion-ready reels on the fly.

05

Music & Artists

320% lift

Studio sessions, interviews, behind-the-scenes — distilled to the moments that drive catalog streams.

06

Brand & D2C

12× ROI

Turn one brand film into 200 paid-media variations. Hook analysis built for premium organic reach.

PRICING

Pay per clip, not per seat.

Start free. Upgrade when you ship. Cancel anytime — your clips stay yours.

Starter

$0free forever
Start clipping
  • 60 min / month
  • 10 clips
  • 9:16 export
  • Community support
Most popular

Pro

$19per month
Try Pro free for 7 days
  • 600 min / month
  • Unlimited clips
  • All aspect ratios
  • Claude hook calibration
  • Priority queue

Elite

$99per month
Go Elite
  • Everything in Pro
  • Brand templates
  • API access
  • Dedicated support

Need 10+ seats, an API, or a custom brand deal? See the full plans →

ready when you are

Stop clipping. Start shipping.

Paste a YouTube link. Get 10 ready-to-post clips in 5 minutes.
No card. 7-day Pro trial. Cancel anytime.

7-day Pro trialSelf-hostableOpen test set