Challenges
RIS d.o.o. as a technology company specialized in business software development, operates with numerous users who send daily inquiries, technical questions and requests for support. Although the company already uses classic customer service tools, such as e-mails and phones, an increasing number of queries – many of which are repetitive and routine – place a heavy burden on technical staff.
This condition results in prolonged response time to more complex demands, reduced efficiency and the risk of user dissatisfaction. In addition, there is a need for better internal organisation of the processing of user requests, better availability of information and greater degree of automation in order to reduce reaction time and ensure a high level of service.
Solutions
In order to respond to these challenges, RIS d.o.o. has joined the program Test Before Invest in cooperation with the EDIH Adria team. The aim of the test was to examine the possibilities of using advanced AI-based solutions, in particular with the help of large language models (LLM) and the Retrieval-Augmented Generation (RAG) technology.
Through the testing process, a functional prototype of a digital assistant was developed, which simulated real user scenarios in a limited environment. The assistant was integrated with a small knowledge base of the company, including FAQs, instructions for using the software and examples of user queries, and demonstrated the following functionalities:
- Automatic classification and prioritization of incoming emails
- Search for relevant content from internal documents
- Generate suggested responses that can be approved or edited by the agent
- Interface for visualization of communication and review of previous interactions
Particular attention is paid to data security and the preservation of the privacy of business information, which is extremely important for RIS given the sensitive nature of user data.
Results and Benefits
Testing of the prototype has brought numerous insights and showed concrete benefits that could be achieved by implementing such a solution in everyday work. The most important benefits include:
- Significantly Reducing the burden on customer service, since the AI system could automatically download a large part of routine queries
- Accelerating the response to users, in particular for simple and repetitive information and technical guidance
- Increasing Customer Satisfaction, because the system would be available 24/7 and allow for prompt responses
- Greater internal efficiency, as employees can devote more time to more complex technical challenges and product development
- Consistency of response, since the AI system always uses the same validated sources of knowledge
- Solution scalability, which can be easily adapted to more users and new scenarios without a proportional increase in the number of employees
Lessons learned
The project has shown that the implementation of advanced AI systems in customer support is not only a technical challenge, but also an organizational one. Key lessons include:
- Scope of application to be defined in advance – AI assistant works best in clearly structured domains with predictable queries
- The quality and up-to-dateness of the knowledge base is crucial for the accuracy of the response
- Initially, it is necessary to involve people as supervisors who can edit answers before sending, in order to avoid inaccuracies
- It's important interoperability with existing support systems, such as CRM tools, user bases and ticket tracking tools
- Users must be aware of the possibilities and limitations of AI assistants, in order to facilitate their use
Measurable data
Based on testing and simulations, the following potential impacts were assessed:
- Reducing the burden on employees by 50-60% in query processing operations of lower complexity
- Shortening the response time by more than 70% on typical, repetitive queries
- Reduction of the total number of escalations towards higher technical support by up to 40%
- Increasing support availability to 24/7, without the need for additional staff recruitment
- Improving the accuracy of responses Linking with validated documents and reducing subjective errors
Conclusion
Through collaboration with EDIH Adria team, RIS d.o.o. has successfully tested an advanced AI solution that can revolutionize the way customer support is provided. Although it is a prototype, testing has shown clear potential for transforming business processes, increasing operational efficiency and strengthening customer relationships.
The following steps include a detailed analysis of the functional requirements, the development of technical documentation for the development of the production version of the assistant and the definition of a strategy for its integration into everyday work. With this project, RIS d.o.o. confirms its focus on innovation and uses artificial intelligence to strengthen competitiveness in the digital economy.
English 
Croatian 
Italian