LLM consulting increases conversion and reduces returns
An AI shopping assistant combines product and behavioral data, promotes appropriate purchasing decisions, and reduces returns.
NEW: With the AI Shopping Assistant, you can offer LLM-based advice – for exciting purchasing decisions and a lower return rate
Did you know that...
65%
of e-commerce companies plan to implement LLM-based personalization solutions (Source: BytePlus, 2025)
27%
of LLM usage (the largest share) in 2024 was held by the retail/e-commerce sectors (GlobalNewswire, 2025)
30%
and more of API growth through 2026 will be driven by tools with LLMs (source: Gartner, 2024)
21%
of companies already using GenAI productively – an increase of 10% in 4 months (Source: Gartner, 2024)
The AI Shopping Assistant provides dialogue-based advice during the consideration phase.
The AI Shopping Assistant heralds a new era in e-commerce: personalized advice in online stores – just as if an experienced sales assistant were standing right next to the customer. This is made possible by a large language model (LLM) that processes natural text and voice input in chat – including combinable filter and sorting functions. The AI Shopping Assistant is integrated directly into the results list, offering maximum flexibility, seamless integration, and context-sensitive advice in real time.
Especially during the important consideration phase, it accurately identifies the needs of your online shoppers and filters out the right products from your range. It combines current product data with the click and purchase behavior of online shoppers to make recommendations even more precise and personalized.
This reduces uncertainty, promotes the right purchasing decision, and lowers the return rate. Your customer receives exactly the advice they need, finds the perfect product, and makes a purchase with confidence. That's exactly what Epoq Advise offers you.
During the consideration phase, potential customers actively compare different products in order to prepare for a purchase decision.
Take a closer look at how our AI Shopping Assistant works
Our AI Shopping Assistant combines modern technology with an intuitive shopping experience. Six clearly structured functional areas demonstrate how it makes digital shopping smarter, faster, and more personalized – without disrupting users' familiar behavior. The functionality can be fully integrated into existing processes and offers your online shoppers real added value. The familiar product search remains unchanged – online shoppers can decide for themselves at any time whether they want to use the AI Shopping Assistant or navigate in the traditional way. The whole thing is demonstrated using the example of a wine shop – but our AI technology is flexible and can be used for all industries and product groups.
1. Smart Start: How the AI Shopping Assistant works right from the start
Link to product search and autosuggest
The AI Shopping Assistant is integrated directly into the product search. An icon in the search box indicates that personal assistance is available – clicking on it starts the consultation. This icon also appears in autosuggest next to matching suggestions, so online shoppers can access advice directly as they type – quickly, intuitively, and at just the right moment.
Free choice: classic search or personal assistance
Whether classic search or intelligent advice – online shoppers decide for themselves how they want to proceed. Clicking on the icon activates the AI Shopping Assistant and takes you directly to the search results list with the assistant integrated. Those who prefer to search without assistance can use the regular product search as usual – without any restrictions.
Advice on category pages
Optionally, the AI Shopping Assistant can also be integrated into category pages. When a specific category is clicked on, the bot appears in the header in a context-sensitive manner – this way, the advice begins directly in the appropriate subject area, supported by the online shopper's previous interaction.
2. Smart Conversation: From the start of the conversation to product advice
Conversation starters to help you get started
When you first start using the AI Shopping Assistant, it suggests suitable conversation starters – such as "Dry Rieslings from Germany" or "Italian white wines compared." These suggestions are based on product data and previous click and purchase behavior, and are dynamically generated by our AI engine. Once you get started, they no longer appear, keeping the experience focused.
Natural language communication and product lists
As soon as a question is asked, the assistant responds in real time with a short, helpful answer in natural language and a product list relevant to the question. The behavior of online shoppers in the store – such as previous clicks, products viewed, or purchases – influences the weighting and selection of the products suggested.
3. Contextual Understanding: How consulting becomes truly intelligent
Understanding complex queries (semantics)
The AI Shopping Assistant also understands complex queries with multiple criteria – for example: "I'm looking for a red wine from Southern Europe between $20 and $30." It recognizes intentions and restrictions and formulates appropriate recommendations – for example: "Sure, I'd be happy to show you red wines from Southern Europe in the $20 to $30 price range. Have fun discovering Mediterranean wines!"
Colloquial language and error detection
Even with unclear or incorrect input, the response remains accurate. Spelling mistakes such as "Chadonney" or colloquial language (such as "Sis" instead of sister) are interpreted correctly, as are regional dialects ("Hasch du Päzer Woi?").
Adaptation to language style and user behavior
The assistant adapts to the language of the online shopper. Those who use informal language are addressed informally, while those who use formal language are addressed formally. Brand-specific tones can also be defined. In addition, click and purchase behavior can be used for better interpretation – for example, to respond more specifically to individual preferences and suggest suitable alternatives.
4. Intelligent Results Lists: Dynamic, explanatory, familiar
Dynamic tags and full filter and sort function
Tags such as "dry" or "semi-dry" are displayed with the results list and help with quick filtering – without any new input. These dynamic tags can be combined with the classic filters and sort function, which remain fully usable – for a familiar user experience. When hovering over products, additional tags such as vintage, growing region, or grape variety are displayed. Clicking on them filters the results list directly, allowing for even faster navigation.
Diverse results instead of limited suggestions
The results lists can show more than just five suggestions – unlike many chatbots, our bot is directly integrated into the results list, providing enough space for product lists. Products with complementary relevance can also appear, tagged with explanatory tags such as "whiskey glasses" for the search term "whiskey."
Familiar experience, intelligently enhanced
Click and purchase behavior is factored into the weighting for displaying the results list. This produces personalized, context-related results – without replacing familiar behavior, as with classic search.
5. Content Discovery: Answers to all questions about the product
Content instead of dead ends
The AI Shopping Assistant not only answers questions about products, but also about general topics – such as "Where can I find your store?" or "What is your legal notice?" The answers link directly to relevant pages, e.g., a blog, an information page, or a service page.
Intelligently linked content
This turns product searches into thematic orientation. The assistant recognizes when content rather than products are required and automatically suggests suitable articles or help content.
Recommendations beyond products
Personal preferences are also taken into account: depending on click and purchase behavior, the assistant recommends, for example, themed worlds, promotions, or advice content. This creates a seamless shopping experience with real added value – beyond the product itself.
6. Goal achieved: From the hit directly to the product page or into the shopping cart
Once the user has found the right product, it can be added to the shopping cart directly from the results list or explored further on the product detail page – without media discontinuity or context change.
The AI Shopping Assistant is therefore not a detour, but a shortcut to the right product – seamlessly integrated into the familiar purchasing process.
Dialog-based advice directly in the online shop – simple, fast, tailored
"The AI Shopping Assistant combines modern LLM technology with our AI engine, which we have been developing for 20 years, and supplements classic search in online stores with dialogue-based advice. Users can search in the same way they are used to with ChatGPT – in natural language, even for abstract questions. The system understands complex relationships and provides suitable product recommendations – for a flexible, hybrid search experience that better suits the diversity of today's user needs."
Stephan Kölle | CTO | epoq internet services GmbH
Our AI Shopping Assistant offers you these features
Context
The AI Shopping Assistant also understands complex queries with multiple conditions and restrictions. This allows users to search precisely, for example, "wine from Southern Europe, but not white wine, under $20," and still receive relevant results.
Reasoning
The AI recognizes connections between search terms, user intentions, and preferences. Instead of just responding to individual keywords, it thinks for itself and provides relevant product recommendations that match the overall desire.
Error Tolerance
The assistant also interprets typos, colloquial language, and regional dialects correctly. This ensures that user questions remain understandable and are answered reliably, even if they are not perfectly formulated.
Language Adaptation
The assistant automatically recognizes whether users are informal or formal and adjusts the tone accordingly – for a natural and customer-oriented approach.
Brand Tone
In addition, the message can be adapted to the desired brand style so that communication always matches the brand identity.
Memory Function
The AI Shopping Assistant recognizes returning users and remembers preferences or previous conversations in order to continue the consultation in a meaningful way and make it even more personalized.
Voice Search
In addition to traditional text input, voice search offers a convenient and modern alternative for finding products using voice commands.
Conversation Starter
The AI Shopping Assistant offers smart introductory questions that are tailored to the situation, season, or target group, making it easier to start a conversation and actively guiding users through the store.
Ranking
In addition to product data, the click and purchase behavior of the respective online shopper is taken into account when displaying the results lists, so that the most suitable products are displayed in real time.
Result Set
The results lists show significantly more than the usual five suggestions. Products with only partial relevance can also be displayed, ensuring that the selection remains diverse and transparent.
Dynamic Tags
Intelligent tags such as "dry" or "under €20" allow users to quickly filter results without having to retype anything. When hovering over a tag, additional tags such as vintage or growing region appear – one click adjusts the list directly.
Classic Filters / Sorting
In addition to the dynamic tags, all the usual filters, sorting options, and other shopping elements remain fully functional and combinable.
Content
The assistant not only provides product recommendations, but also relevant information such as FAQs, service pages, and advice texts – for comprehensive guidance.
Merchandising
Based on user behavior, relevant topics, promotions, and campaigns are displayed to enrich and personalize the shopping experience.
Shopping Cart
Once the user has found the right product, it can be added to the shopping cart directly from the recommendation – without any media disruption or change of context.
The AI engine of our AI Shopping Assistant: LLM meets smart personalization
Our AI engine is at the heart of the AI Shopping Assistant. It combines the power of LLMs for natural and contextual speech output with your store's individual data – such as your product catalog, customer click and purchase behavior, and your domain knowledge. This creates an intelligent knowledge base that delivers accurate product recommendations while responding naturally to queries.
Since the AI Shopping Assistant is a module of our personalization software, the knowledge base that has already been generated can be seamlessly used for other modules. This allows for flexible personalization of additional touchpoints in the store. This combination of state-of-the-art AI technology and data-driven personalization ensures a unique, individually tailored shopping experience.
Learn more about our AI Technology.
Check out our insights on digital consulting
For many years, we have been supporting online shoppers with digital advice – initially in the form of guided selling and visual filters. This expertise has been incorporated into the further development of the AI Shopping Assistant. The result: an intelligent, dialogue-based shopping companion that provides personal support directly in the product search.
01 / 4
Merchandising control: Our AI Shopping Assistant follows your business logic
Our AI Shopping Assistant adapts not only to your customers, but also to you. In addition to automatic, AI-based product recommendations, you can incorporate your own specific requirements. You can control the results list using integrated merchandising options. The combination of intelligent personalization and manual control makes the AI Shopping Assistant a highly flexible sales tool – tailored to your product range, your strategy, and your KPIs.
Prioritize & Push
Determine which products, brands, or categories should be displayed preferentially and highlighted specifically.
Business Objectives
Define which business objectives – such as inventory clearance, margin strength, or new products – should be taken into account when displaying products.
Rules
Define which rules should apply to specific use cases or target groups.
Customer Targeting: Our AI Shopping Assistant also thinks in terms of target groups
The AI Shopping Assistant doesn't just think in terms of products – it also thinks in terms of target groups. You decide who gets to see what: personally, segmented, or according to strategic priorities.
1:1 Personalization
Recommendations are based on the behavior of each individual user – in real time. This results in personalized advice that really fits.
Segment-based Targeting
Target specific groups: new customers, frequent buyers, bargain hunters, or those interested in seasonal items – everyone receives relevant content and product recommendations.
Product-based Targeting
Want to put the spotlight on specific products? The AI Shopping Assistant connects your business goals with the right target group – automatically and controllably.
Ready to go in 3 steps – this is how the integration of the AI Shopping Assistant works
Generation of a knowledge base
In order to tailor the AI Shopping Assistant optimally to your products, we need your product catalog in XML or CSV format, as well as data on the click and purchase behavior of your shop customers. We collect this behavioral data via a tracking code that you simply integrate into your online shop.
01
Integration into your shop system
We will discuss your visual ideas together and send you the div element as a placeholder for the front end of the AI Shopping Assistant. You can then integrate this into your shop system at the desired location.
02
Live consultation start in the shop
Once everything is integrated, your customers will have access to personalized, AI-powered advice – perfectly embedded in the customer journey of your online store.
03
Monitoring and control of the AI Shopping Assistant
The Epoq Control Desk allows you to monitor and control various KPIs of your digital product advisor.
Key Figures
In the "Key Figures" tab, you can view various KPIs for your digital product advisor: click performance, sales rate, conversion funnel, response times, and session duration.
When is it worthwhile for you to use an AI Shopping Assistant?
Consulting
Are your online shoppers unsure which product is right for them? The AI Shopping Assistant proactively supports them in their selection – in real time, personally, and around the clock.
Bad Buys
You are experiencing high returns in certain product categories. These are often caused by false expectations. Our assistant recommends suitable products based on needs, thereby reducing mispurchases from the outset.
Complexity
Does your product range require intensive consultation or explanation? The AI Shopping Assistant helps make products understandable and supports your online shoppers in their purchasing decisions.
Want to help your online shoppers make the right choice?
With our AI Shopping Assistant, you can leverage technological advances to reduce returns and increase customer satisfaction.
Popular blog articles on digital product advice
01 / 8