In our projects, we are increasingly incorporating solutions based on artificial intelligence. A great example is the interactive assistants we’ve developed to make using applications easier. A modern AI assistant is much more than a simple chatbot — it’s a fully integrated component of the application that not only answers questions but actively supports the user by guiding them step by step through available features. This way, when someone returns to the app after a break and doesn’t remember how to perform a specific task, the assistant not only gives hints but also points to the relevant interface elements and intuitively walks the user through the process, making the application much more accessible.
On the technical side, we rely on retrieval-augmented generation (RAG) and the ability of language models to call functions. The entire solution is implemented using the Semantic Kernel library, which simplifies integration and development. This approach offers great flexibility — it allows us to update the assistant’s knowledge by editing documentation and expand its capabilities by implementing new tools the assistant can use.
The possibilities are broad — from simply highlighting interface elements, through step-by-step guidance, to automatically executing selected actions in the app.
Such solutions help reduce the burden on support teams by handling the most common questions (“Hey Damian, how do I generate a report?”), while giving new users a stress-free, confident entry into the app. Importantly, they don’t replace humans but truly support their work. We believe that in the future, assistants like this will be a natural extension of applications — combining traditional GUIs with the flexibility of language models. The user will simply choose whether to click or describe what they need.
