Buying-intent scoring
A cheap keyword check first, then AI scoring only where it can matter.
Before any AI call, a word-boundary keyword pre-filter drops posts that do not really match. That keeps cost down and keeps unrelated text away from the model.
Text that passes gets a content score: a sentiment score and a buying score. A cheap model scores first. The expensive model runs only when the buying score crosses a threshold and your budget allows it.
Scores are cached by a hash of the text, so the same text is never scored twice. When your monthly AI token budget runs out, scoring stops until the next month; nothing fails.
- Scores
- Sentiment and buying score per text
- Model tiers
- Cheap first; expensive only above the threshold
- Budget
- Monthly AI token budget per organization