This article investigates the potential of large language models in overcoming problems within the business negotiation process. Typical issues encountered in business negotiations at various communication stages, which highlight the necessity for systematic resolution (such as breaches of good faith principles, ineffective communication strategies, lack of clear planning, etc.), are identified. The study focuses on a key negotiation problem related to the increasing transaction costs of businesses due to unmet interaction results. It demonstrates how large language models can optimize the negotiation process at every stage: from preparation (gathering and analysing information, scenario modelling) to conducting negotiations (contextual analysis, proposal formulation) and implementing agreements (execution control, monitoring). The study highlights that large language models can ensure: objectivity and speed in information processing; improved planning and forecasting; enhanced communication efficiency; and reduced transaction costs. The findings of the research conclude that the application of large language models is a promising development direction for optimizing the business negotiation process, contributing to the reduction of transaction costs and strengthening companies’ competitiveness by expanding human capabilities in managing business communications.