Collecting functional quantitative intra matrix data in experimental samples of articular cartilage is still challenging due to its delicate complex heterogeneous structure in which constituents are intermingled right up to the ultramicroscopic level. Any attempt to insert a transducer inside this material via piercing would damage the structure leading to unrepresentative data. Traditional non-invasive methods are technically difficult for obtaining precise functional data. This paper presents a novel computational approach, using the agentbased concept, to create a ‘virtual microscope’ that can be used to provide functional information throughout a heterogeneous complex medium, such as articular cartilage, in silico. The method involves two-dimensional cellular automata, a hybrid agent, new local agent rule and a traditional neighbourhood rule. The hybrid agent combines constituents of the system (solid and fluid) where the local rule determines intra-agent evolution. The proposed approach was validated by simulating diffusion into a model of cartilage matrix that was characterized with anisotropic permeability. The simulated results were then compared to magnetic resonance imaging (MRI) data. Spatial map of diffusion at different times and depthdependent diffusion profiles were provided in colour-coded pictures. Qualitative and quantitative comparison of results with experimental data shows that this novel approach can accurately and efficiently represent diffusion of fluid into the cartilage matrix. It demonstrates the potential of hybrid agent and local rule to enhance agent-based techniques for porous materials and other areas of research. We conclude that the ability to establish a “virtual microscope” offers a viable opportunity for in-silico experiments that can extend our knowledge beyond the capability of traditional laboratory experiments, while also facilitating information for creating models for numerical methods such as finite element analysis, meshless and smoothed particle hydrodynamics. The combination of the approach presented here with conventional simulation methods can provide a framework for modelling and analysis of complex porous materials. We concluded that the hybrid agent and local rule concept introduced in this paper can also be potentially exploited to enhance many of the existing agent-based techniques.