FDA TEMPO Pilot Adds Four Digital Health Companies

The U.S. Food and Drug Administration has confirmed a four-company opening cohort for its TEMPO digital-health devices pilot, expanding from Dexcom’s late-July start to three additional participants by 21 August 2026. DistilINFO reported the lineup on 4 September: Dexcom’s Glucose Health Program, SonderMind’s app for adults 22 and older with moderate depression or anxiety, Limbic’s Unpacked AI voice CBT for Medicare patients, and Cadence Solutions’ HypertensionOS for Stage 2 hypertension medication management.

Four tools spanning glucose to hypertension

TEMPO expands to Technology-Enabled Meaningful Patient Outcomes for Digital Health Devices. According to DistilINFO’s TEMPO cohort report, Dexcom pairs real-time glucose data with AI-driven insights aimed at prediabetes and Type 2 diabetes screening and control. SonderMind’s smartphone app is designed to sit alongside therapy or medication for adults with moderate depression or anxiety.

Limbic’s Unpacked product uses an AI voice agent to deliver cognitive behavioral therapy to Medicare patients with significant depression or anxiety. Cadence Solutions’ HypertensionOS helps physicians and advanced practice registered nurses manage and adjust blood-pressure medication for adults with Stage 2 hypertension. Together the four tools sketch the pilot’s early clinical map: metabolic sensing, behavioral health, and cardiometabolic medication titration.

• Dexcom Glucose Health Program — glucose data plus AI insights
• SonderMind app — adults 22+ with moderate depression or anxiety
• Limbic Unpacked — AI voice CBT for Medicare depression/anxiety
• Cadence HypertensionOS — Stage 2 hypertension med management

Generative AI under FDA discretion

Both Cadence and Limbic rely on generative AI. The FDA still lacks formal guidance specifically for generative-AI-backed devices, and DistilINFO reported that the agency will use discretion on requirements such as premarket authorization while reviewing these tools through the pilot. That makes the cohort a live policy laboratory as much as a product validation exercise.

The timing overlaps a broader FDA conversation: a generative AI-enabled medical devices discussion paper with public comments due 19 October 2026. The paper is not final guidance, but it signals how risk assessment, premarket evaluation, and postmarket monitoring for GenAI devices may evolve. TEMPO participants generating conversational or adaptive outputs will likely become informal reference cases as that docket fills.

ACCESS model and a path to forty

Participating manufacturers must collect, monitor, and report real-world performance data for intended uses. Devices are expected to be covered and used by patients in CMS’s ACCESS model, the July-launched payment design meant to expand technology-supported chronic care for Medicare beneficiaries. Linking regulatory pilot status to reimbursement pathways is deliberate: clinical evidence and payment evidence are supposed to move together.

The FDA plans to select up to 40 TEMPO participants across four clinical use areas — cardiometabolic risks, cardio-kidney-metabolic diseases, chronic pain, and behavioral health. Four companies against a forty-slot ceiling means most of the program’s selection story is still ahead. Health-system product teams watching how frontier models enter regulated workflows can also track OpenAI GPT-6 Astra AGI-era launch and adjacent creative-tool interface shifts in Midjourney v8.2 Edit Lightbox alpha.

For hospital innovation offices, the near-term checklist is concrete: map which chronic-care pathways already sit inside ACCESS; ask vendors whether their GenAI stacks fall under TEMPO-style real-world reporting; and treat FDA discretion as a temporary bridge, not a permanent free pass. For digital-health founders, the signal is that generative interfaces are entering regulated chronic-care pilots before a finished GenAI device guidance document exists — so evidence plans should assume evolving expectations through at least the October 2026 comment window.

File this as health-science news: a four-company TEMPO start, two GenAI participants under discretion, a CMS ACCESS tether, and a stated ceiling of forty slots across four disease clusters. Outcomes from Cadence and Limbic may quietly shape how the agency later writes formal rules for generative clinical tools.

Real-world performance reporting is not optional window dressing for TEMPO. Manufacturers must monitor how devices behave in intended use — including failure modes that lab studies miss, such as hallucinated CBT suggestions, incorrect titration advice, or biased screening thresholds. Clinicians should demand clear escalation paths when an AI voice agent or medication assistant is uncertain.

Because ACCESS ties payment to technology-supported chronic care, TEMPO outcomes may influence which digital tools Medicare beneficiaries actually receive. That dual track raises equity questions: will rural or low-bandwidth patients get the same agentic CBT or hypertension OS experiences as urban clinics? Pilot dashboards should track access gaps alongside clinical endpoints.

Device makers outside the first four should treat the open slots as a signal to prepare evidence packages now — usability, adverse-event logging, GenAI change-control, and human oversight design — rather than waiting for final GenAI guidance. Commenting on the October 19 discussion-paper docket is a parallel path to shape the rules that may eventually replace today’s discretion.

Sources

• 4 Companies Join FDA’s TEMPO Pilot — DistilINFO, 4 September 2026
• FDA generative AI-enabled medical devices discussion paper feedback — FDA, comments due 19 October 2026