The onset of damage and its evolution in polymers at the nanoscale (<100 nm), as well as the mechanics that govern it, have not been well-studied. Recently, atomistic simulations have revealed that there is a stark difference between the mechanical behavior of materials at the nanoscale and macroscale. In particular, how local chemical heterogeneities: influence damage nucleation and growth at the nanoscale, alter the makeup and mechanical behavior of the fiber/matrix interface and interphase regions, and manifest themselves at the macroscale. Given that interphase volume fractions can exceed 50 vol. % in a nanocomposite, it has been hypothesized that small changes at the interface and within the interphase may yield drastic changes in bulk behavior. Recent attempts at answering these questions have primarily involved electron microscopy, but unfortunately, polymers are inherently non-conductive and commonly experience charging and electron-induced damage when imaged. AFM can be used as an alternative to electron microscopy. Specimens can be readily molded and tested at ambient conditions, without the need to put under vacuum or sputter coat. Nevertheless, AFM has its own challenges, which include the time required for sample imaging, tip selection and calibration. In this study, carbon nanofibers (CNFs) were decorated with silver (Ag) nanoparticles via a solventless process and subsequently introduced into a commercial polyurethane (PU) to promote the formation of a conductive network. Four sample types were fabricated: neat PU, PU+CNF, PU+CNF+Ag Acetate, and PU+Ag-CNF. Nanofiller concentrations were held at 3.47-6.78 % vol. for CNF, 0.06-1.31 % vol. for Ag Acetate, and 0.01-0.27 % vol. for Ag. High-resolution SEM, and AFM were used to observe the morphology of freeze fractured, and uniaxially strained specimens, respectively, while TEM and nano X-ray computed tomography provided two-dimensional and three-dimensional views of nanofiller dispersion in each system, respectively. DSC and TGA provided Tg and nanofiller concentrations in the composites, respectively. Chemical maps of sample surfaces were acquired via AFM-IR, while the onset and evolution of damage (prior to plastic deformation) was monitored via in situ tensile testing (quasi-static) in AFM coupled with DMT modulus mapping. By using DMT modulus, rather than height or phase signals, the components and damage states within the composite can be clearly differentiated in situ. The ability to conduct these experiments in situ and at ambient conditions provides an unprecedented amount of insight into the influence that chemical/morphological heterogeneities have on the nanomechanical behavior of polymer nanocomposites, which is essential for guiding and refining current computational model frameworks.