AI Recommendations Engine: Beta Now Open

After 6 months of private alpha with 50 partners, AI Recommendations is now open to all NexaTech customers.

How it works

The engine combines collaborative filtering (what users like you clicked) with content-based signals (item embeddings from our AI Text API) in a real-time ensemble model.

Cold start

New users are served content-based recommendations until enough behavioral data is collected (typically 5 events). This is configurable.

A/B testing built-in

Every recommendation experiment is an A/B test by default. The dashboard shows CTR, conversion rate, and revenue impact segmented by experiment arm.

Integration

A JavaScript snippet handles event tracking. The recommendation endpoint returns ranked item IDs with explanations. Total integration time: under 30 minutes.