After Kauz.ai with its aiSuite was selected as the best solution provider, the phase of technical design and implementation began. Initially, only a website chatbot for the Warendorf District Health Department was planned. However, since the website content was integrated into the district administration’s global website and therefore difficult to separate clearly, the use case was expanded—at the initiative of the project sponsor, the Health Department—to include additional departments.

The no-code aiStudio enables the methodical and rapid implementation of customized AI chatbots. A key prerequisite for this is high data quality, which was ensured by the existing state of the service portal and helped accelerate the project implementation. Thanks to the AI architecture based on the RAG approach (Retrieval-Augmented Generation), new content can continuously be incorporated without the need for extensive model training.

A comprehensive website chatbot ensured a seamless user experience for website visitors. In order to continue offering a well-functioning AI chatbot beyond the end of the project, the data must be updated regularly and ongoing dialogues need to be reviewed for accuracy. The aiStudio provides the appropriate no-code tools for this purpose, such as a dashboard and simple data import workflows.

While the first chatbot primarily served a support function, the new AI chatbot was to be developed into a navigation assistant for the “Mein ias” portal. To leverage the benefits of artificial intelligence, the existing chatbot had to be completely replaced. Conceptual groundwork, such as structured question lists, could be reused. However, an in-depth examination of the extensive portal content was necessary to clearly distinguish non-AI-eligible topics, such as medical advice. For instance, while the AI chatbot is not allowed to provide a diagnosis for pain, it can offer tips for promoting mental health.

The content and technical integration of the chatbot followed a well-defined project methodology. Solutions had to be developed for technical challenges, such as integration with live chat for non-AI-eligible topics and differentiation between content accessible before and after the website login barrier.

The AI chatbot was operational within a few months without any technical issues. Using the Conversation Viewer in aiStudio, ongoing chatbot dialogues are regularly analyzed internally. When needed, additional content can be uploaded, and dialogue plans can be independently extended by ias experts to ensure controlled AI usage. In its first month alone, the chatbot recorded an average dialogue length of six questions and answers per user, indicating high user acceptance of the AI chatbot.

The potential for further AI chatbot applications in other business areas is being centrally evaluated as part of the ias Group’s portfolio management and represents a key element of their AI roadmap. The development of another chatbot for the webshop has already been approved.

The technological maturity of artificial intelligence and augmented reality in terms of technical feasibility is almost 100% today. However, the challenge is to establish the technological solutions as a service and sustainable business model in the sense of a platform approach. A key use case within the research project is, for example, the scanning of heating system nameplates and the creation of a database that provides tradespeople with self-help information via their AR glasses or smartphone. Manufacturer-related stand-alone solutions are only of limited help here, as repair work is carried out on different and sometimes unknown systems. It is therefore necessary to develop a centralised, cross-provider knowledge database that can be used by many user companies. It is therefore not just a question of technological implementation, but above all of designing a collaborative ecosystem in which costs and benefits can be distributed in a value-adding manner.

Through cooperation with an electrical, painting and plumbing, heating and systems company, the work processes and information requirements of the tradespeople and customers were recorded in detail and compared and prioritised with possible AI support. Kauz.ai supports domain-specific knowledge management with the aiStudio and provides the user interface for targeted information transfer with its AI assistants. ODAV - as the leading technology partner for German trade businesses - hosts the AI solutions centrally so that no technical maintenance work is required at user companies.

All user companies in the Minerva project were equipped with simple digital assistants within a very short space of time, be it product consulting integrated into the painting company's website or various tools for the independent optimisation of energy purchasing and the associated electrical installations. Further expansion stages of process digitalisation are being worked on, always with the focus on ensuring that other trade businesses can also benefit from the solutions.

The research project will run for three years with the aim of establishing practice-orientated research work for various trades as a concrete market offering. An important focus is the promotion of the digital mindset in the skilled trades. The Minerva project group provides suitable solutions for businesses that are ready for change, which will continue to be operated by the commercial project partners in the long term. The added value of rapid adaptation of AI in the skilled trades is enormous and ranges from up- and cross-selling potential through 24/7 availability, cost savings through the reduction of travelling times to solving the shortage of skilled workers through the successful integration of migrants with the help of multilingual AI assistants.

The positive experiences with chatbots have gradually led to an expansion of automated communication, which should be taken even further and implemented in multiple languages using generative AI. The variety of topics and the number of trade fairs require further development of the content by creating synergies between the individual chatbots.

The content and organisational complexity of the chatbot deployment is stringently managed by the scalable multi-client system. All chatbots access a central database for general information and thus benefit from the content enhancements of each individual chatbot, which can be flexibly activated or deactivated. This approach also applies at configuration level: every trade fair chatbot benefits from improvements in prompting and model adaptations. Documents can be uploaded or links stored to answer specific questions. Thanks to the analytics features, the internal project team knows exactly which dialogues are taking place and which content areas need to be optimised and expanded in order to guarantee a satisfactory quality of information.

The use of chatbots has already led to a reduction in the volume of e-mails and thus a reduction in the workload. Trade fair visitors from other time zones now also have a reliable source of information at all times. Content adjustments can be implemented independently and easily within the company using aiStudio. If questions or challenges nevertheless arise, the Kauz.ai Customer Success Team is characterised by a fast response time. Thanks to the platform approach, Westfalenhallen automatically benefits from developments in the AI market, as these flow directly into product development. New trade fair chatbots can now go live within a few weeks on average.

The Group's chatbot infrastructure is constantly being further developed through the event and trade fair programme and the experience gained. The positive acceptance of chatbots, which has once again increased significantly due to the boom in ChatGPT, is driving the further expansion of communication automation in the Westfalenhallen with Kauz.ai as a technology partner.

In recent years, VR-NetWorld has expanded the use of chatbots to 32 partner banks. Through this collaboration, various AI methods and product variants from Kauz have been employed to achieve the goal of a digital customer journey with high service automation for frequently recurring customer inquiries. By combining high standardization of chatbots with flexibility for content customization, both large and small banks were able to effectively use chatbots for customer communication.

With the advent of generative AI at the end of 2022, the capabilities of chatbots fundamentally changed. Suddenly, chatbots exhibited significantly improved conversational abilities. If large language models could be successfully combined with proprietary content, chatbot content expansion could be accelerated, reducing the manual effort needed for chatbot maintenance.

Kauz.ai was one of the first chatbot providers worldwide to integrate its technology with ChatGPT, migrating its clients to the new technological standard within a short timeframe. This began with a hybrid chatbot implementation, where the existing chatbot infrastructure was gradually enhanced with ChatGPT-generated responses and tested in parallel operations for optimization. With the launch of aiStudio—Kauz's GenAI framework—new chatbots, such as the one for VR-NetWorld's service portal, were developed.

Thanks to Kauz.ai’s swift adaptation to technological advancements, VR-NetWorld was able to enhance the conversational capabilities of its deployed chatbots and gain crucial experience with generative AI within just a few months. In addition to improving chatbot quality, operational effort for chatbot maintenance was also reduced.

As an early adopter of customer chatbots, VR-NetWorld successfully leveraged the potential of generative AI quickly and effectively. The use of chatbots has gained significant momentum within the corporate group and is set to expand on an even larger scale.