Over the past decade, there had been progress in the development of cell therapy for insulin-dependent diabetes. Nevertheless, important hurdles that need to be overcome still remain. Protocols for the differentiation of pluripotent stem cells into pancreatic progenitors or fully differentiated β-cells have been developed. The resulting insulin-producing cells can control chemically induced diabetes in rodents and were the subject of several clinical trials. However, these cells are immunogenic and possibly teratogenic for their transplantation, and an immunoisolation device and/or immunosuppression is needed. A growing number of studies have utilized genetic manipulations to produce immune evasive cells. Evidence must be provided that in addition to the expected benefit, gene manipulations should not lead to any unforeseen complications. Mesenchymal stem/stromal cells (MSCs) can provide a viable alternative. MSCs are widely available from many tissues. They can form insulin-producing cells by directed differentiation. Experimentally, evidence has shown that the transplantation of allogenic insulin-producing cells derived from MSCs is associated with a muted allogeneic response that does not interfere with their functionality. This can be explained by the immunomodulatory functions of the MSC subpopulation that did not differentiate into insulin-producing cells. Recently, exosomes derived from naive MSCs have been used in the experimental domain to treat diabetes in rodents with varying degrees of success. Several mechanisms for their beneficial functions were proposed including a reduction in insulin resistance, the promotion of autophagy, and an increase in the T regulatory population. However, euglycemia was not achieved in any of these experiments. We suggest that exosomes derived from β-cells or insulin-producing cells (educated) can provide a better therapeutic effect than those derived from undifferentiated cells.