LLMs in e-commerce – successful with artificial intelligence

LLMs (Large Language Models) have experienced a real hype. Especially since ChatGPT came onto the market, LLMs have been on everyone’s lips – including e-commerce companies, who are now asking themselves how they can position themselves and what advantages they can gain for their business. We examine the current models on the market and present our own embedding model, FisMwAsp, specifically for German-speaking countries.

Ecommerce IT ‐ 12. June 2024

LLMs − What does that mean

Large language models, also known as AI language models, are used to understand human language and generate content. They are trained with huge amounts of data and are constantly “learning” through continuous training. LLMs can be described as problem solvers and help, for example, with answering questions, summarizing content and creating texts.

They are often used in customer support in the form of chatbots, and ChatGPT from Open AI also provides support with useful functions in many areas.

In short: LLMs facilitate communication between humans and technology; artificial intelligence created by humans for humans. The possible applications are diverse and depend heavily on the respective business context. Below we provide a brief overview of three common applications in e-commerce and how they are used in practice.

LLMs in use – focus: e-commerce

Customer Service

A classic: large language models in customer service.

LLMs are used there to operate chatbots and voice assistants that can answer customers’ questions about the company, its processes and products.
In addition to answering questions, the chatbot can trigger all kinds of processes – whether orders, search queries, customer reviews or registrations. This allows customers to talk to a system that understands natural language and can respond to their needs.

Content generation

LLMs can also be used for product descriptions, marketing texts and automatically generated, personalized emails. This helps companies to save significant resources in content creation while producing consistent and SEO-optimized content. For example, an LLM can be used to create appealing and informative product descriptions from a list of product features.

User content validation

In e-commerce, the validation of content by users – such as ratings and comments – is an intensive time and cost factor. LLMs can be used to automatically check content for inappropriateness, spam and fake reviews.
Request advice now

Examples – How LLMs can be used

Amazon Alexa and Amzon.com

Amazon uses its voice assistance technology Alexa not only for home automation and personal assistants, but also to improve the shopping experience on Amazon.com. Customers can use voice commands to search for products, place orders and check the status of their deliveries. This significantly improves the customer experience through a seamless and interactive shopping process.
More information: www.amazon.com

Alibabas AliMe

Alibaba has developed AliMe, an intelligent digital assistant that is able to respond to customer requests in real time. AliMe is not only used for customer service requests, but also to provide them with interactive and personalized shopping experiences. The assistant can guide customers through the entire shopping process, make recommendations and help with questions about products.
More details: d3.harvard.edu

Medienwerft and the German market

The use of LLMs in the German e-commerce market is nowhere near as mature as in the English-speaking world. This is made particularly clear by the small number of embedding and language models that speak German.
The FIS Group consisting of FIS, Medienwerft and FIS-ASP is therefore particularly proud of the development and publication of the FisMwAsp embedding model, which has been trained to better process German technical language. In this way, we were able to make a small contribution to the community.

FisMwAsp – concentrated expertise from FIS, Medienwerft and FIS-ASP combined

Thanks to the group’s internal development, we were able to combine expert knowledge from various specialist areas to create a particularly powerful and versatile German language model.

To ensure the security of internal company data and prevent its unauthorized disclosure to third parties, we paid strict attention to GDPR compliance during development. Thanks to internal LLM hosting by FIS-ASP, we counter this risk and enable companies to retain full control over their data.

As an SAP Gold Partner, we specialize in seamless integration into SAP systems, which enables rapid implementation and optimized processes. Of course, other systems can also be easily connected to our model, allowing companies of various scales, whether B2B or B2C, to benefit.

This synergy within the FIS Group offers our customers everything from a single source, tailored to their needs and without unnecessary frills. We are continuously developing our solutions to meet the requirements and technological developments of tomorrow.

AI chat, content generation or AI-supported process optimization: where do you want to start?

Let us advise you now!

    Get in touch with us

    All fields marked with * are required for ordering and processing. Your person-related data will be used by us solely for the purpose of processing your inquiry according to our privacy policy.


    Find out everything important about Medienwerft – experts in brands & ecommerce for over 25 years – here:

    About us


    „Behind every impressive online solution there is a well thought out technological concept. My team of system analysts, database experts, front-end developers, back-end professionals and experienced designers ensure that everything runs smoothly. Feel free to talk to us.“

    Frank Meier

    Managing Director

    Tel: +49 40 / 31 77 99-0
    Email: info@medienwerft.de