We’re inviting providers, payers, employers, and other care delivery organizations to study what happens to outcomes and costs when their populations get AI-enabled behavioral health support.

Why we’re creating this program

Over the past year, conversations with health plans, risk-bearing provider groups, employers, and other groups have kept converging on the same question: what would Ash do for our population—our diabetic patients, our heart failure patients, our members stuck on a six-week waitlist for a therapist, our members whose repeat panic attacks land them in the ED? Until now, we’ve explored those questions one partner conversation at a time. With this program, we’re opening a dedicated partnerships channel, with the goal of facilitating evidence generation in AI for mental health alongside organizations that take responsibility for people’s health: payers, payviders, risk-bearing primary care groups, and care delivery organizations to study what happens to outcomes and costs with AI behavioral health support.

What we’ve built, and what we know so far

Slingshot’s mission is to help a billion people change their minds and lives in the ways that they want to. Towards that mission, we’ve trained our foundation model purpose-built for psychology and behavior change and built Ash—the first AI designed for mental health, used by over 500,000 people.

The real-world evidence we’ve collected so far shows:

• Symptom improvement that replicates. In a 305-person study with NYU, Ash users showed significant improvements in depression and anxiety, a finding that replicated in a dose-response study of 100,000 users, where more engagement with Ash also predicted greater improvement.

• Gains beyond symptoms, in the measures that matter to patients. The NYU study found improved social support, such as more time with others, and more calls to friends and family. In a separate cohort of 1,284 new users, sleep quality improved 14.7%, relationship satisfaction 11.6%, and life satisfaction 10.2% over the first four weeks of use.

• Early signals suggesting cost reduction. In self-reported health care utilization data, Ash users with chronic medical conditions (e.g., diabetes; cardiovascular disease) were 60% less likely to have repeat urgent-care visits (9.5% vs. 23.3% of non-users), reported ~40% lower absenteeism, and had higher medication adherence (92% vs. 76% taking medications as prescribed). These are exactly the signals we want to pressure-test against a partner’s claims data.

These are observational findings, and we’re careful about what we claim from them. What we are working on demonstrating now, with partners like you—who have a defined population and the data to match—is what this does to claims, utilization, and disease-specific outcomes.

Who this is for

Organizations that take responsibility for people’s health: payers, payviders, risk-bearing primary care groups, employers, and care delivery organizations. The two things that matter most are genuine interest in exploring what AI can do in new areas of health, and access to the data that answers the real questions — claims, utilization, outcome measures.

What we bring:

• We fund Ash access for the study population

• Our research team is hands-on from design through last patient out, eager to collaborate with our partners on cohort definition, endpoints, engagement strategy, IRB navigation, and other components of research

• We have evidence generation infrastructure that has been used to run real-world studies at scale, and a commitment to disseminating what we find

What we’re looking for:

• Partners with genuine interest in exploring what AI can do in new areas of health.

• Access to data that answers the questions about improved lives and costs (claims, outcome measures)

Populations we’re most interested in

We believe in a holistic model of health, recognizing that behavioral health plays a strong role in the management of chronic disease and is a modifiable driver of cost. Populations of particular interest to us right now include: comorbid depression and anxiety in heart failure, type 2 diabetes, medication adherence across chronic disease, COPD, chronic pain and MSK, and care transitions (like stepping down from intensive outpatient care). With that, we’re open to pilot studies in other areas of need identified by our partners in healthcare delivery organizations—ultimately, we want any work with Ash to be solving a real pain point for patients and clinicians.

What a pilot looks like

Scope is case-by-case, but a typical structure would include: a defined cohort within your population, Ash access funded by us, jointly defined endpoints spanning clinical outcomes, utilization, and cost, a 6-12 month observation window, and findings both parties can use. We move from first conversation to launched pilot in months, not years.

Start with a short questionnaire

Tell us about your population and the question you’d want answered. We don’t need a formal proposal or procurement packet; we review on a rolling basis and follow up to schedule a call.

Questions first? Reach out to research@slingshotai.com, and we’re happy to talk feasibility before you fill anything out.

We look forward to working with you!

The Slingshot Research team