02Predict
Hear it before
it gets loud.
Societal forecasts which ideas in your niche are about to break, how likely they are to go viral, and how long you have before they go stale.
Forecast scope · niche: Food & CulinaryHorizon 72h · CI 80%
Viral probability
82%
Peak in
+31h
Window
47h
Saturation
Low
Related sounds & tags
02.1How the forecast is made
Four stages, from raw posts to a probability.
- IN01/04
Ingest
Public posts across your niche, refreshed continuously and deduplicated across platforms.
- EMB02/04
Embed
Captions, transcripts and visuals mapped into one vector space, so the same idea is recognised in any format.
- CLU03/04
Cluster
Similar posts grouped into trends. Every trend tracks its velocity, acceleration and spread.
- FCST04/04
Forecast
A model trained on past breakouts estimates the chance of going viral, the peak time and the decay.
02.2Rising now
The watchlist.
| Trend | Niche | Viral prob. | Peak | Stage |
|---|---|---|---|---|
| Street-food POV, 1-take | Food | 82% | +14h | Rising |
| Ingredient fact-check | Beauty | 79% | +9h | Breaking |
| AI tool, 1 minute test | Tech | 74% | +22h | Rising |
| Hidden-spot coordinates | Travel | 71% | +30h | Early |
| Salary-day breakdown | Finance | 63% | +2d | Early |
02.3Specifications
- Horizon
- 6 hours to 7 days, per trend
- Outputs
- Viral probability, peak time, window length, saturation, related sounds and tags
- Signals
- Velocity, acceleration, creator spread, cross-platform echo, comment intent
- Models
- Embeddings by OpenAI, reasoning by Claude, time-series on Vertex AI
- Honesty
- Every forecast shows its confidence band. Low-confidence calls are labelled as such.
Early access · Batch 01