Injectable protein or peptide drugs now constitute ~10% of the pharmaceutical market. Recently, we developed a cell-based method to stably deliver protein drugs. To do this, we coupled CRISPR/Cas9-based nucleases with adeno-associated virus for delivery of donor homology templates to safe-harbor loci in human B cells, which we subsequently differentiate into antibody-secreting cells (ASC). A subset of these cells resembled long-lived plasma B cells (CD38hiCD138+), whereas others exhibited phenotypes (CD38+CD138-) not previously associated with longevity. We show that engineered B cells can engraft in recipient immune deficient, NOD/SCID/gc-null (NSG) mice and stably produce human antibody for >1 year. ASCs engineered to express firefly luciferase primarily migrated to the bone marrow, the endogenous location of human long-lived plasma cells. Upon provisioning NSG mice with human cytokines (IL6, and/or BAFF) that promote survival of long-lived plasma cells, we observed substantial increases in antibody production and durability of B cell grafts. BAFF preferentially promoted class-switched, CD138+ plasma cells. In contrast, IL6 promoted surface IgM-expressing CD38+CD138- ASCs, as well as CD138+ plasma cells. Finally, quantification of human ASC in the murine bone marrow and spleen showed that as few as 20,000 engineered cells/recipient was sufficient to maintain IgG titers of 10 ug/mL, levels that could be of therapeutic valuable if achieved using expression of candidate mAb reagents. Together, these studies show that engineered human B cells have the capacity to engraft long-term and function normally in vivo, strongly supporting further studies using this novel cell therapy platform for long-term delivery of protein drugs.