<p>Most metaheuristic algorithms rely predominantly on information provided by the best-performing individual to guide their search strategies. However, this approach often leads to rapid dissemination of its influence throughout the population, reducing diversity and diminishing the algorithm's ability to explore alternative regions of the solution space, ultimately causing premature convergence. In contrast, models such as cellular automata introduce state modifications based on local neighborhood interactions, influencing the states of nearby cells, whereas agent-based models (ABM) establish interaction rules among agents that regulate collective behavior. This paper proposes a novel metaheuristic approach that considers collective information from individuals by integrating the principles of cellular automata and ABM. This method utilizes dynamically defined neighborhoods, comparing the quality of adjacent individuals against a threshold of feasible solutions that are adapted based on population diversity. This approach employs three specialized mechanisms: an escape movement for poor-performing neighborhoods, a decreasing sine movement to refine promising solutions, and a transverse orientation movement when the global best is within the neighborhood. The proposed approach was benchmarked against 30 test functions encompassing multimodal, unimodal, and hybrid problems, and compared with eight well-known metaheuristics. The results demonstrate that the proposed method delivers a competitive performance, showing robust consistency and faster convergence for solving complex optimization problems.</p>