An AI influencer-matching engine
Matches in seconds instead of about seven hours
- Sector
- Marketplace and marketing technology
- Market
- United Kingdom
- Timeline
- 2 weeks
Secondsto match, against about 7 hours by hand
Top 2matches consistently accurate
2 weeksfrom kickoff to a working engine
The challenge
Matching influencers to campaigns was manual and slow, about seven hours per campaign, with low accuracy, budget lost to trial and error, and trust undermined by fake followers.
What was built
- Onboarding for campaigns and influencers
- Audience and content-context analysis across posts and videos
- Relevance matching on historical performance
- Real-time matches that combine several data layers
How it was built
- A vector-based relevance engine on PostgreSQL with pgvector
- Generative and contextual content analysis