Meet ARIEL: the AI shopping assistant that doesn't follow a script
We're officially launching ARIEL, a brand new AI shopping assistant of a new generation for online stores.
We've been working on ARIEL for a long time, testing and improving it with real shoppers in real stores, and we're proud to finally introduce it properly. Here's what it does, why we built it the way we did, and where it’s heading next.
We didn't want to build another chatbot
You've probably met the classic e-commerce chatbot. It pops up in the corner, offers you three buttons ("Delivery", "Returns", "Talk to us") and does fine as long as you stick to the questions someone predicted in advance. Ask it something slightly different and it melts down: "Sorry, I didn't understand. Did you mean: Delivery options?"
That's because scripted chatbots don't understand anything. They follow decision trees. Every answer has to be written, every path has to be mapped, and the moment a shopper steps outside of it, the conversation ends.
"We didn't want to build another one of those. Shoppers don't come to a store with a script. They come with a need, often a vague one, and they deserve someone who can work with that."
Michael Zelinka
ARIEL reasons instead of reacting
ARIEL doesn't follow a decision tree. It reads what the shopper actually wrote, works out what they need, and answers using the store's real catalogue, prices, business rules and knowledge base.
That means it can:
understand needs in the shopper's own words, even when they don't know what the product is called
find and compare the right products, and explain why they fit
ask a clarifying question when it genuinely helps, instead of guessing
answer practical questions about delivery, returns or the store's terms, and hand over to the store's team when a human is needed
There are no synonym lists, decision trees or hardcoded rules to maintain. When the store updates its catalogue or content, ARIEL picks it up automatically.
From guessing product names to solving problems
For now, ARIEL lives right in the search bar, or just below it. And that changes what searching means.
Until now, search worked only if the shopper guessed the approximate product name: "family board game 2–4 players easy rules"
With ARIEL, they can simply describe their problem: "Something fun to play with my grandparents that doesn't take forever to learn"
ARIEL understands what they're after, finds games that fit, and explains why each one works for that evening. Shoppers stop hunting for the right keyword and start finding solutions.
Made to sound like your store
ARIEL represents the store and its brand. Each store sets her name, tone of voice and opening message, so it sounds like the brand rather than like a generic AI. It answers from the store's own knowledge base and sells the store's own products.
Shoppers can write to her in any language, and the store stays in control of when and how a conversation is handed over to a person.
Implementation is quick and needs minimal IT involvement. For teams that want deeper control, an API is available too.
ARIEL’s just getting started
This launch is the beginning, not the finish line.
ARIEL will keep getting smarter and gaining new skills. We want ARIEL to help the way a great shop assistant does: explaining which product fits your situation, how to use it, and what else you’ll need to get the job done. But here’s the thing: it’s still just one small piece of what we’re building for 21st-century eCommerce. We’ll share more soon!
In the meantime, we’d love to show you what it can do with your own catalogue.
Raventic tým
