Health & ScienceHealth & Science 5 min read

FDA TEMPO Pilot Lets AI Care Tools Reach Patients Early

The U.S. Food and Drug Administration’s TEMPO pilot is putting selected digital health—and in several cases AI-enabled—chronic care tools in front of patients before those intended uses receive

PC

PromptCrates Editorial

Staff Writer

0 0
FDA TEMPO Pilot Lets AI Care Tools Reach Patients Early

The U.S. Food and Drug Administration’s TEMPO pilot is putting selected digital health—and in several cases AI-enabled—chronic care tools in front of patients before those intended uses receive full marketing authorization. On the agency’s public participant table, manufacturers including Limbic, Cadence Solutions, SonderMind, and Dexcom appear alongside clinical use areas such as behavioral health and cardio-kidney-metabolic care. The program’s core trade is explicit: earlier access under CMS Innovation Center ACCESS participation, paired with mandatory real-world data collection, while the FDA notes effectiveness has not yet been evaluated for the pilot uses.

How TEMPO pairs FDA discretion with ACCESS

TEMPO stands for Technology-Enabled Meaningful Patient Outcomes for Digital Health Devices. According to the FDA participant page, the agency intends to exercise enforcement discretion for certain requirements—such as premarket authorization and investigational device requirements—when a selected manufacturer’s device is offered to or by CMMI ACCESS participants for an intended use expected to be covered under that model. ACCESS, short for Advancing Chronic Care with Effective, Scalable Solutions, is the CMS Innovation Center vehicle that makes the care pathway operationally relevant for Medicare-aligned populations.

The first TEMPO participant announcement landed on 22 July 2026, and the table has continued to expand. That timeline matters for clinics and digital health founders watching generative and agentic tools enter regulated care: TEMPO is not a free-for-all sandbox. It is a named list with intended uses, contraindications, and an explicit disclaimer that pilot effectiveness remains unevaluated. Manufacturers must monitor and report outcomes whether they operate as ACCESS participants themselves or supply devices to other ACCESS organizations.

For health systems outside the United States, TEMPO is still a signal. Regulators from the EU to ASEAN watch how the FDA handles AI voice agents and protocol-bound titration software when evidence is incomplete. The U.S. choice—conditional early access plus data duties—will influence how payers and ministries design their own adaptive pathways, even when local statutes differ.

Which AI-heavy participants are on the list

Limbic Inc.’s Unpacked is among the clearest AI entries. The FDA describes it as an AI-voice agent delivering structured cognitive behavioral therapy-based programs inside outpatient behavioral health services for Medicare beneficiaries with clinically significant depression and/or anxiety. Intended benefits include increased access to structured psychological treatment, measurement-based care, and real-time safety flagging for clinical oversight. The contraindication list is long and strict, spanning suicidality, psychosis, dementia, eating disorders, substance use as a primary condition, pregnancy, frailty, and non-English speakers, among others.

Cadence Solutions’ HypertensionOS targets clinician-supervised, protocol-bound antihypertensive medication initiation and titration for adults with diagnosed Stage 2 hypertension when eligibility criteria and deterministic safety checks are met. It is intended for licensed clinicians with prescriptive authority rather than direct-to-consumer self-dosing. SonderMind’s Adjunctive Care Application aims to reduce depression and anxiety levels for adults 22 and older as an adjunct to psychotherapy and/or pharmacotherapy via smartphone, with its own exclusions around mania, psychosis, PTSD, and suicidality.

Dexcom’s Glucose Health Program sits in early cardio-kidney-metabolic and CKM use areas, combining monitoring, tailored guidance, and AI insights to support screening for prediabetes and type 2 diabetes and potentially improve glycemic control. Unlike a single diagnostic algorithm authorization, the program framing emphasizes ongoing behavioral and metabolic support inside ACCESS-aligned care. Separately, PromptCrates covered the distinct FDA Queen of Hearts STEMI ECG clearance this morning—TEMPO is a different regulatory story about chronic-care access, not De Novo authorization for acute ECG triage.

What clinicians and builders should watch

Clinicians considering TEMPO-linked tools should read intended uses and contraindications as hard product boundaries, not marketing footnotes. An AI voice CBT agent that excludes non-English speakers or complex psychiatric comorbidity will fail silently if intake workflows are sloppy. HypertensionOS assumes protocol-bound clinician supervision; treating it like an unsupervised chatbot would miss the safety model the FDA summarized.

Builders and investors should notice the evidence bargain. TEMPO can accelerate patient contact, but it also locks companies into real-world data operations that resemble continuous clinical study infrastructure. Logging, outcome monitoring, and safety flagging are not optional extras—they are the price of enforcement discretion. Generative and agentic features will face extra scrutiny as the FDA’s separate August discussion paper on generative AI-enabled devices solicits comments through mid-October 2026.

Patients and caregivers should understand the difference between pilot access and proven benefit. TEMPO devices may reach Medicare-aligned users earlier, yet the FDA is clear that effectiveness for those pilot intended uses has not been evaluated. Asking clinicians how outcomes will be monitored—and what happens if the pilot ends—is a reasonable patient safety question, not a rejection of innovation.

TEMPO will not replace traditional authorization pathways, and it should not be blurred with device clearances like Queen of Hearts. It is a narrow bridge between chronic-care payment experiments and digital tools that increasingly rely on AI. The winners will be teams that treat real-world evidence as a product requirement from day one, not a slide for later.

Sources

FDA TEMPOLimbic UnpackedCadence HypertensionOSDexcomdigital healthACCESS model

Related articles