Tufts researchers have put a dollar figure on what agentic AI can do for clinical trials. A new analysis from the Tufts Center for the Study of Drug Development (CSDD) finds that an AI clinical monitoring agent can add up to $21M in net financial value to a single drug development program, with returns as high as 82 times the investment in late-stage trials.
The study evaluated Medable’s Clinical Monitoring Agent using expected net present value modeling, benchmarked oncology data from Tufts CSDD, and contract value data from Medable’s cloud platform. Expected gains reached roughly $7.5M for a Phase II trial, $11.3M across combined Phase II and Phase III development, and $21M for a Phase III program. Return on investment came in at 64x for Phase II and 82x for Phase III.
Most of the value comes from operations rather than the model itself. Tufts estimated direct savings of about $4.4M per Phase II and $5.6M per Phase III study in on-site monitoring, plus $600K to $1.7M in administrative off-site monitoring tasks. Fewer site visits, lower travel costs, and faster enrollment and database lock timelines drove the gains, according to Tufts CSDD executive director Ken Getz.
Getz said the work marks the first time such modeling, based on actual use and benchmark data, has quantified the financial impact of an agentic AI tool in drug development. For sponsors weighing AI investments, the analysis offers a concrete baseline for what autonomous trial software can return, and it shifts the conversation from adoption cost to operational efficiency.
