The article is dedicated to the implementation of the Alt Linux technical support assistant based on LLM technology. The prospective architecture of the system is described in details. It is also noted that the prospective architecture of the system is associated with massive data exchange between LLM models, which requires significant computational resources. Therefore, the rough draft architecture is proposed, which has considerably lower resource requirements. The Retriever module aimed at Alt Linux documentation preprocessing and indexing is considered. The experiments on choosing and evaluating the LLM model are discussed, as a result the Qwen3-32B model is chosen for further integration into the system. The assistant user interface is described in details. It is noted that the Qwen model turned out to be a very successful solution for the assistant due to the ability to use the “normal” and reasoning modes when answering a question. It also can be stated that modern LLM models have reached a level that allows them to be successfully used in intelligent dialog systems. As the future direction, authors plan to implement a prospective system architecture using more advanced hardware and to increase the amount of documentation on Alt Linux that will be used by the assistant.