FAQs

Before the era of generative AI, most chatbots were rule-based, i.e. they worked according to predefined rules and decision trees. The use of NLP, i.e. extensive vocabularies, was necessary for natural conversations. The effort required for larger areas of application for chatbots was correspondingly high. With generative AI, both the design of conversation rules and the creation of the necessary vocabulary can be reduced by up to 90 %. The performance of AI chatbots has made a quantum leap compared to their predecessors. Hybrid chatbots combine the rule-based approach with artificial intelligence, further increasing the accuracy of chatbots.

Chatbots are traditionally used to automate customer service. Website chatbots are available around the clock and can answer simple, repetitive customer queries independently. In an expansion stage, they can be used for further workflow automation such as the assignment of internal service tickets. While an interaction with a service employee costs between 6 and 12 dollars, the costs with chatbots fall to 0.50 to 1 dollar.* The new possibilities of artificial intelligence will also expand the field of application of chatbots in practice. Chatbot solutions for internal knowledge management or the optimization of IT help desks are becoming increasingly important.

*99 Firms: Chatbot Statistics

The biggest risk when using chatbots is customer dissatisfaction. If incorrect answers are generated or the customer service journey does not allow for forwarding from the chatbot to the human in difficult cases, the quality of service decreases. It is therefore important that companies create positive experiences with their chatbot deployment and are able to identify and immediately rectify shortcomings in automated service communication. It goes without saying that compliance with data protection regulations and the legally compliant use of artificial intelligence must be guaranteed.

The success of the use of chatbots can be measured using a variety of key performance indicators (KPIs). The most important KPIs include:

• Accuracy rate: Percentage of answers that are correct and satisfactory.
• Error rate: Percentage of answers that are incorrect or unhelpful.
• Engagement rate: Percentage of users who interact with the chatbot compared to the total number of visitors.
• Total resolution rate: Proportion of user queries that are fully resolved by the chatbot without the need to be forwarded to a human employee.
• Escalation rate: Percentage of conversations that need to be forwarded to human employees.

We are one of the few platforms to offer hybrid chatbots, i.e. to further increase the potential of generative AI by integrating individual vocabularies and ontologies. In contrast to ChatGPT agencies that customize the LLM for individual applications, we provide customers with a sustainable platform solution that transparently maps chatbot interactions and supports the independent further development and management of chatbots through an easy-to-use user interface for prompt engineering features and various LLM connections.

We offer chatbot solutions in various levels of complexity. Website chatbots for small and medium-sized enterprises can go live within two weeks.

We currently offer a chatbot-as-a-service for energy suppliers. Historically, we have many years of experience in chatbot development for insurance companies and financial service providers, and an as-a-service offering is being planned.

Yes, chatbots for product advice can also be easily configured with the aiStudio. Multimedia product information can be added to the corresponding texts in the Multimedia Center.