AI recommendations

Personalized product suggestions powered by Vertex AI Discovery Engine.

Track behavior

await window.genmb.commerce.track('view-item', {
  userId: 'u_123',
  documents: [{ id: 'product_42' }],
});

Get recommendations

const { data } = await window.genmb.commerce.recommend.personalized({
  userId: 'u_123',
  count: 5,
});

Event tracking runs in a 10-second Redis-buffered batch. Don't worry about call volume.

Recommendations fall back to popular-products when Discovery Engine isn't available, so the SDK always returns something usable.

FAQs

What signals power recommendations?▾

GenMB forwards user events (view-item, add-to-cart, purchase-complete, search) to Vertex AI Discovery Engine. The more events, the better the personalization.

How long does Discovery Engine need to learn?▾

It serves cold-start recommendations from day one. Personalization quality improves continuously with traffic; meaningful personalization typically starts after a few hundred events per user.