About
Taste is hard to search. We fixed that — for the whole store.
Food & beverage shoppers arrive with taste-based intent — "smoky mild for tacos", "low-caffeine evening tea" — that default Shopify search can't handle. bevvi turns that intent into grounded, cited recommendations, and gives merchants visibility into what shoppers actually want.
Always grounded
Every claim is cited from your catalog and content.
Never guesses
On allergen or medical questions, it defers — it never invents safety.
Fair economics
Predictable per-conversation pricing, no surprise AI bills.
Our thesis
One engine. Many niche funnels.
Generic AI search treats every store the same. We do the opposite: one grounded recommendation engine underneath, marketed through the language of each vertical it serves. Coffee shoppers hear about roast and brew; hot-sauce shoppers hear about heat and smoke; tea shoppers hear about mood and caffeine — same platform, funnels that actually speak their category.
That's why we launch narrow — coffee, hot sauce, and tea first — for founder-led Shopify food & beverage brands with nuanced catalogs and real gifting demand. Prove conversion lift in a niche, then bring the same engine to the next one.
What we ask your store for
Installing an app means handing it permissions, and Shopify files ours under headings that sound broader than what we do with them. So here is the whole list, what each one is for, and what we do not do with it.
View store data: products, Online Store
read_products
Your catalog is what the assistant recommends from. Titles, descriptions, images, prices, tags and stock come across so a shopper gets a real product with a real price instead of a guess.
- Nothing is written back to your store.
- Your catalog is never used to train a model, and never shown to another merchant.
View staff and contributor data
read_content
Your pages and blog posts are how the assistant answers questions about shipping, returns and how to use what you sell. It quotes your own copy rather than inventing an answer. Shopify counts this as staff data because articles carry an author name.
- Author names are not stored, shown to shoppers, or used for anything.
- No staff accounts, permissions or contact details are read.
We do not ask for order or customer records. Nothing we read is used to train a model, and nothing crosses between stores.