A new FDA pilot is giving some generative AI health products a route to market before the agency has finished authorizing them.
STAT reported September 3 that Cadence and Limbic both have AI products in the first cohort, with four devices accepted into the TEMPO pilot so far. Companies in the program can put their software in front of patients without first obtaining marketing authorization. In exchange, the agency gets a live look at how each tool performs outside the lab.
TEMPO feeds a parallel Medicare experiment called ACCESS, which tests payment for technology that helps beneficiaries manage chronic conditions. A larger device pool gives that experiment more options to evaluate. The pilot also hands FDA reviewers exposure to cutting-edge AI in real-world settings, experience that should carry over when similar products arrive through formal review.
Generative AI is straining the agency’s premarket framework. Software that can change how it decides as it learns does not fit cleanly into the fixed design reviews the FDA built its device program around, so the pilot trades some premarket certainty for direct observation.
Limbic is an AI mental health platform used by parts of the NHS, and Cadence builds software for chronic care management. Live use data gathered under TEMPO could inform later authorization decisions for both.
The FDA has not said when the remaining TEMPO participants will be named or how long the pilot will run.
