Українська ❤️ 
Русский 
E-commerce case studyJediDesk AI × Horoshop
AI beauty advisor for cosmetics: the TANOYA and JediDesk case study
Cosmetics presales — the consultation before a purchase — requires particular care. The advisor needs to understand the shopper's goal, skin type, desired result, and where a product fits into a routine. JediDesk AI asks follow-up questions and recommends TANOYA products using the shopper's answers and current catalog attributes rather than a manager's personal preferences. It also handles routine questions about availability, payment, delivery, and orders 24/7.
About TANOYA: a catalog where guidance shapes the purchase
TANOYA manufactures and sells professional skincare products for the face and body. In this category, shoppers often choose an entire care routine rather than one product. Before the purchase, the team needs to ask the right questions, explain the differences between products, and still answer routine questions about availability, payment, and delivery.
The team needed an AI beauty advisor for its online cosmetics store that could work on two levels: provide attentive product recommendations based on shopper needs and deliver reliable routine support 24/7.
Why cosmetics recommendations require attentive presales
The quality of the pre-purchase conversation directly affects how confident a shopper feels about their choice. Routine support should not take the team's time away from meaningful guidance.
Recommendations require follow-up questions
Good guidance starts by clarifying the shopper's goal, skin type, desired result, and intended use.
Purchase intent does not wait for business hours
If guidance is delayed until the team is online, the shopper may not reach a confident decision while their interest is high.
343 products and many criteria
The request must be matched quickly with product purpose, properties, price, and current availability.
Before the AI advisor
- A manager reconstructs the context and asks the same basic questions each time.
- A recommendation may depend on one person's experience and preferences.
- Questions about availability, payment, delivery, and order status interrupt detailed presales guidance.
After JediDesk AI launched
- AI asks basic follow-up questions in the same logical sequence.
- Recommendations use the shopper's needs and facts from the current catalog.
- AI handles routine support and hands specialist or unusual questions to the team.
How JediDesk AI combined cosmetics recommendations with routine support
The assistant was connected to the store chat and trained on TANOYA data. It leads the first stage of the conversation: gathering context, matching the request with the catalog, and explaining the differences between products. At the same time, it answers questions about availability, payment, delivery, and orders, then hands requests that need expertise to a person.
Current catalog
The assistant can use 343 products, descriptions, prices, and availability during the conversation.
Rule-based recommendation
AI suggests products using shopper answers, catalog attributes, and configured rules rather than a manager's personal preferences.
Routine support 24/7
It answers questions about availability, payment, and delivery, and returns the current order status.
Smart escalation
The assistant does not replace a specialist: sensitive, unusual, and wholesale questions go to the responsible manager.
How TANOYA prepared data for AI recommendations and support
The direct JediDesk AI integration with Horoshop gives the assistant current attributes, prices, availability, and order data. Recommendations therefore follow the shopper's request and catalog data rather than memory or personal preference.
For technical setup guidance, read How to install JediDesk chat on Horoshop.
How the AI assistant recommends cosmetics
The assistant clarifies what the shopper needs, checks the answers against the product feed, and suggests relevant products using catalog attributes. It also uses current prices and availability, answers routine service questions, and hands complex requests to a manager.
Clarifies the need
Asks structured questions about the request, skin type, and desired result.
Reads the product feed
Uses descriptions, categories, properties, prices, and current availability.
Builds a recommendation
Selects options against the criteria and explains which product properties were considered.
Brings in the team
Hands the context to a manager when a request is complex or needs a specialist's decision.
How the AI advisor guides shoppers from need to choice
These real scenarios show both levels of the experience: AI provides pre-purchase guidance, asks follow-up questions, and explains catalog-based recommendations; it also resolves routine service questions or brings in a manager.
No customers' personal data is published. Photo scenes are editorial reconstructions; screenshots show excerpts from the assistant's original Ukrainian-language conversations.
Cosmetics recommendations for a skin type
The shopper was looking for SPF and products for dehydrated skin. AI clarified the request step by step, found matching items, and explained the differences. The recommendation used the shopper's needs and product data rather than a consultant's personal opinion.
2 products matched to the shopper's criteria
Excerpts from a real JediDesk AI conversation in Ukrainian; information that could identify the customer is not shown.
Support as a reliable foundation
After receiving product guidance, a shopper should not have to wait for a simple service answer. The assistant explains payment and dispatch times, checks orders, and completes a routine conversation on its own.
Conversation resolved by AI
Excerpt from a real delivery conversation in Ukrainian.
Order status check
In a private chat, AI finds an order in Horoshop by order number or phone and returns the current status.
Handing complex requests to a manager
When a request concerns wholesale cooperation or needs a specialist's decision, AI does not invent an answer; it hands the context to a manager.
Handed to a manager
Choosing a set for a skincare routine
AI clarifies the desired result, suggests relevant sets, and explains the contents and purpose of each option. It does not call a product “perfect” without supporting data and brings in a manager when the request goes beyond the knowledge base.
Guidance based on catalog data
AI advisor results in the first month
JediDesk AI handled more than a third of all customer conversations, including product guidance and routine service requests. This is an overall chat metric, not a presales- or sales-only result; it is reported separately to avoid overstating AI's influence on product choice.
of requests handled by AI
The share of requests in which the assistant handled the inquiry.
of conversations resolved by AI
Conversations completed without an operator.
scenarios in the knowledge base
Trained answers for common shopper situations.
Methodology. The figures cover the assistant's first full month and come from the TANOYA case statistics in JediDesk. “Handled by AI” and “resolved by AI” are different metrics and should not be treated as interchangeable.
How AI recommendations and routine support help the team
AI does not replace expert guidance. It standardizes the start of a presales conversation, matches the request with catalog data under configured rules, and handles service routines, leaving the team to focus on situations that genuinely need a person.
A consistent start to every conversation
Every shopper receives the same basic follow-up questions, regardless of the team's workload.
More time for complex recommendations
Managers join unusual questions instead of repeating the basic flow and routine service answers.
Clear recommendation boundaries
The assistant uses the catalog and hands the conversation to a person when professional assessment is needed.
How to launch AI recommendations and support in an online store
Explore the product overview, browse available integrations, or see how an AI advisor works in a website chat.
Launch your AI assistant in one day
Connect your store's product feed and JediDesk AI will use names, descriptions, categories, prices, and availability to recommend products for each shopper's request. In the same chat, the assistant will answer routine questions and hand complex conversations to your team.
- Create an accountGet access to your JediDesk workspace.
- Connect the product feedGive AI the current data from your catalog.
- Launch recommendations and supportAI advises shoppers and hands exceptions to your team.