Storage and demand flexibility have become increasingly important to keep the electric system functioning with growing shares of renewables, to the point that their contribution can no longer be ignored even in the planning phase, as flexibility can compete with conventional network reinforcement. Since most of the demand flexibility is located at distribution level, the flexibility provided by distribution networks must be considered when planning the transmission network as well. However, the huge size of the models combining both network levels makes the resulting optimization problem hard to tackle even with modern hardware. We propose a novel heuristic procedure for the decoupling of the transmission and distribution grid planning problems based on the formulation of surrogate models of the distribution networks, specifically aimed at capturing the flexibility that can be provided to the transmission network. Using this approach, we obtain a speed improvement of up to two orders of magnitude – increasing with the amount of distribution networks – in solving the planning problem. The heuristic produces good quality solutions featuring a near-optimal objective cost.