NewDays raises $16M for its companion AI against cognitive decline, with evidence from just 24 patients so far

By Zendoric · September 3, 2026.

NewDays, a Seattle startup that combines telehealth with a generative artificial intelligence (AI) companion to treat mild cognitive impairment, has closed a $16 million seed round led by Madrona and General Catalyst. In parallel, the company has announced its arrival in Nevada, the sixth state where it operates, alongside Washington, California, Florida, Texas and New York, according to its own announcement reported by Dealroom.

The product combines two layers: monthly telehealth visits with human clinicians and unlimited conversations with Sunny, an AI companion that guides patients through memory exercises and reasoning games. These are clinically backed techniques, but historically hard to access, according to the source. The stated goal is to use AI to scale an intervention that the shortage of specialists had kept rationed.

The scale of the problem it targets is considerable: one in three Americans over 65 shows some degree of cognitive impairment, with 11% already living with dementia and 22% with mild cognitive impairment (MCI), the earlier stage where early interventions have the most room for impact, according to the data NewDays cites.

The evidence the company presents is still modest. In July, at the Alzheimer's Association International Conference, NewDays presented results from its patients showing performance on cognitive tests above what standard decline curves would predict, equivalent to some 18 months of preserved cognitive function. The study covered only 24 patients, a sample too small to draw firm conclusions, something the company itself appears to accept: it has launched a randomized controlled trial (RCT, the standard for proving that an intervention works and is not a placebo effect or a statistical fluke) together with Kaiser Permanente in California.

On the business side, NewDays lets users try Sunny for free before a 20-minute clinical assessment; from there, each session costs $150. The company accepts traditional Medicare as an out-of-network provider and is working to obtain in-network provider status with both commercial insurers and Medicare Advantage plans. "Generative AI under the supervision of an expert clinician is what finally allows us to deliver a medically proven intervention at that scale. That is the entire thesis of the company," said CEO and co-founder Babak Parviz.

According to Dealroom, the $16 million places this seed round among the top 3% of all health seed rounds recorded in the United States, a sign that investors are willing to bet heavily on AI as a lever to scale a form of care — for dementia and cognitive impairment — that is chronically scarce because of the cost and availability of human specialists.

Our read: the NewDays pattern fits what we have been observing in medical AI that is genuinely advancing. It is not the chatbot that diagnoses on its own, but the narrowly scoped tool working under clinical supervision — the same setup that makes FDA-cleared radiology algorithms or AI-assisted breast cancer screening in the MASAI trial work. Here the human clinician still signs off on the assessment and the plan; the AI supplies the scale — unlimited conversations — that no human therapist could offer at that price.

That said, it is best not to jump to verdicts. Twenty-four patients is a proof of concept, not a demonstration; the very design of the trial with Kaiser Permanente acknowledges that gap. And the business faces the same bottleneck that has sunk more than a few digital health startups in the United States: without broad insurer reimbursement, charging $150 per session limits access for precisely the population — middle-income seniors, many of them dependent on Medicare — that needs the service most. The move to "in-network" status is no minor formality: it will determine whether this is an accessible therapy or a niche product for those who can afford it.

In the long run, however, the attempt points in the right direction of the underlying thesis we keep insisting on: if AI can multiply the reach of already-proven interventions against cognitive decline, one of the greatest burdens of old age stops being a scarce resource reserved for those who can pay for a private therapist. Preserving cognitive function beyond what decline curves predict is, on a small scale and still to be confirmed, a concrete example of how AI can extend years of health and autonomy, not just life expectancy. The RCT with Kaiser Permanente will be the first serious filter for determining whether that promise holds up outside a conference hall.