Rapid urbanization has led to an increase in the number of vehicles on the road. Consequently, quick mobility, one of the basic needs for commuters traveling, is directly hindered due to traffic congestion. The present-day solutions available in the literature suggest re-routing vehicles to alternate paths to avoid crowded locations, which is subject to how the traffic conditions are perceived by a respective system. However, these solutions are not viable for long-term congestion avoidance. Optimizing and improving the conventional traffic management system can relieve traffic congestion. To address the above issue, in this paper, a traffic flow prediction framework called “The Intelligent Traffic Prediction with Dijkstra and LSTM System” has been developed. Firstly, to find the best routes, the system generates adjacency matrices from several network graphs of roads and then applies Dijkstra’s algorithm to the resulting matrix. Next, the accuracy of the route forecasts is increased by constantly updating these matrices to reflect the current traffic conditions. Utilizing AI training, this approach aims to reduce the time taken for path computation without sacrificing accuracy. Furthermore, the technology assists in sustaining a consistent traffic flow, reducing the likelihood of congestion by an average of 87.5% and finding the most efficient routes for individual vehicles.