Existing deterministic methods can solve coupled heat transfer problems in complex 3D geometries, but the increasing complexity results in high computational time and memory usage due to grid-based approximations. This paper presents a semi-meshless method for solving coupled heat transfer problems. This method is based on a stochastic approximation, i.e., the Monte Carlo method. The resulting algorithm is less sensitive to the system's complexity. Thanks to advancements in computer acceleration technology, it is now easier to implement this method. This coupled algorithm was applied to hightemperature porous heat exchanger technology, which could be a potential solution for improved heat recovery. A Kelvin cell structure was used as the porous geometry and placed inside the heat exchanger. The heat recovered in terms of temperature rise was evaluated at different solid thermal conductivity and fluid inlet velocity parameters using the Monte Carlo method and compared to the deterministic method. The salient features of the Monte Carlo and deterministic methods were also discussed.