Researchers demonstrated site‑specific integration of large DNA payloads to program CAR‑T cells directly in vivo, bypassing ex vivo manufacturing. A UCSF team led by Justin Eyquem reported the dual‑vector approach, showing locus‑specific, stable CAR expression and potent antitumor activity in humanized mouse models. Results were described in coverage of the group’s Nature paper and GEN interview. The study combined a targeting vector for T‑cell specificity with a second vector enabling homology‑directed integration at a chosen genomic locus. Investigators reported superior functional persistence and improved control versus random viral integration methods. Tests included aggressive leukemia, multiple myeloma and solid tumor models where in‑body CAR‑T cells mediated tumor regression. If translated, the approach could eliminate weeks of centralized manufacturing, reduce cost and expand access to CAR‑T for patients who cannot wait for cell production. Key technical challenges remain—delivery specificity, off‑target editing risk and regulatory pathway—but the data represent a clear proof of principle for programmable, site‑specific in vivo cell engineering.