Discover how Media Expert & Synerise brought a conversational shopping assistant to the mobile app. Shoppers meet a virtual consultant called Marian, built on Synerise Assistant. It starts every conversation already knowing the person and reads the live catalogue in real time, so the products it suggests are the ones that actually fit and actually exist. Media Expert's launch shows what native behavioral AI infrastructure can unlock when it reaches customers directly.
Subtitles available in English, Spanish, Brazilian Portuguese
Client
Media Expert is CEE's largest consumer electronics and home appliances retail chain (RTV/AGD); it runs a mature omnichannel operation across its stores, MediaExpert.pl, and its mobile app. The app gives customers the full catalogue and their order history in one place. It is where Marian works, turning a live catalogue of millions of products into a guided conversation for each shopper.
The goal
Search boxes make shoppers do the work. Media Expert wanted an AI agent that does it for them: one that opens each conversation already knowing the shopper, reasons over the live catalogue of millions of products as it answers, and hands off to a human the moment a question needs one. It went live mobile-first, across the Media Expert app.
The solution
The solution is built on Synerise Assistant, which runs on Synerise's behavioral AI infrastructure. That is what separates it from a generic AI model dropped into a shop.
A generic model can hold a conversation, but it does not know who it is talking to. Synerise Assistant does, because the behavioral context Media Expert already owns is built into the model's reasoning, in real time and across every channel. Pair that with live product data and a persona tuned to Media Expert's brand voice, and the conversation stops being generic. It becomes Marian.
In the conversation
A shopper opens the app with a rough idea of what they need. Marian greets them based on what they have been browsing or the product they are looking at. It narrows the options with a few quick replies and filters, asks a clarifying question when the request is broad, and explains the trade-offs in plain language. By the end of the chat, the shopper is on the product that fits, not a list they have to sort through themselves.
What Marian does in the app
• Recommends products: similar items, personalized picks, top and most-popular options, cross-sell and complementary accessory categories.
• Opens with a dynamic greeting shaped by what the shopper has been browsing, what they have bought, or the product card they are on.
• Guides the shopper with quick replies and quick filters, asks clarification questions when a request is broad, and adds follow-up prompts in its summaries.
• Looks up order status, either across a customer’s orders or for a single order by number.
• Hands off to a human consultant for questions on payments, delivery, returns, complaints, invoices, order changes or store policy. It checks live-chat availability first.
• Points shoppers to the right place: the mobile app, Media Expert Podróże, and the Konto Plus loyalty program.
• Keeps a chat history, lets shoppers download or email a transcript, and lets them report an issue.
Summary
The assistant is the visible part. The behavioral AI infrastructure underneath is what makes it useful. Because Marian starts every conversation with real behavioral context and reads the live catalogue in real time, it can move a shopper from a rough idea to the right product inside one chat. That is what native behavioral AI infrastructure unlocks.
Big credit to the Media Expert team for taking it live at scale.
