The modern system of industrial risk management at enterprises is based on the basic principles laid down in the standards of industrial safety and labor protection, and presupposes, basically, long-term precedent risk management - responding to their occurrence, analyzing the results, and taking measures to reduce risks. At the same time, with rare exceptions, the management of extremely rapidly developing risk situations, carried out in the course of their development, remains outside the brackets. For example, the failure of some machine tools, or their parts, can be predicted in advance by the parameters of their operation (the nature and level of noise, vibration, temperature parameters, etc.). Previously, the management of such risk situations was complicated by the lack or high cost of automated systems capable of responding with the required delays. The features of risk management of extremely rapidly developing risk situations are considered. A comparative analysis of modern message brokers suitable for building intelligent risk management systems in conditions of uncertainty has been carried out. The architecture of software tools for intelligent risk management in conditions of uncertainty based on the Apache Kafka message broker is proposed. A comparative analysis of the implementation of fuzzy data processing based on Kafka Streams within Apache Kafka and a separate application outside Apache Kafka is carried out, an estimate of the load generated by such software is obtained.