Buoy Health

Buoy Health is an AI-powered symptom assessment tool built in partnership with Harvard Medical School. Designed for everyday consumers, it walks users through their symptoms using a conversational interface and helps them understand what might be causing them, then points them toward the appropriate level of care. It is available free to individual users, with a B2B licensing model for health systems, insurers, and employers who want to embed the tool in their own platforms.

How Buoy Health works

When a user opens Buoy, they describe their symptoms in plain language. The AI model runs a differential diagnosis process, weighing possible conditions against the user’s reported symptoms, medical history inputs, and demographic factors. Rather than returning a raw list of conditions, it guides the user toward a care recommendation: self-care at home, a telehealth visit, an urgent care clinic, or an emergency room. It also provides rough cost estimates for those care options, helping users make informed decisions before they spend money or time on a visit they may not need.

The underlying model was developed with clinical input from Harvard Medical School physicians, which distinguishes it from general-purpose chatbots repurposed for health questions.

Strengths

• Clinical pedigree: The Harvard Medical School collaboration gives the model a credibility foundation that most consumer health apps cannot claim.
• Care path specificity: Instead of stopping at a possible diagnosis, Buoy tells users where to go next and what that choice might cost, which is the most actionable part of the experience.
• Differential diagnosis: The tool considers multiple possible conditions at once rather than anchoring on a single answer, which reduces the risk of dangerous tunnel vision.
• Free consumer access: There is no paywall for individual users, making it accessible to people who cannot afford a doctor visit just to find out whether they need one.

Limitations

• Not a diagnostic tool: Buoy explicitly cannot diagnose conditions. It surfaces probabilities and care recommendations, not clinical verdicts. Users who misread it as definitive medical advice could delay appropriate care.
• Input quality dependency: The output is only as good as the symptoms the user describes. Vague or incomplete inputs can lead to generic recommendations that are not useful.
• No continuity: Buoy does not store health history across sessions for individual consumers. Each interaction starts fresh, which limits its usefulness for tracking symptoms over time.
• B2B focus can dilute consumer experience: As the company deepens its enterprise integrations, the consumer-facing product may not receive the same level of ongoing development attention.

Who it is for

Buoy Health is best suited for people who want a structured, medically grounded starting point when they notice a new symptom. It works well for adults who are unsure whether a symptom warrants a doctor visit, people trying to avoid unnecessary and expensive ER trips, and employers or health plans looking to reduce avoidable care utilization among their populations. It is less suited for people managing ongoing chronic conditions, those who need longitudinal health tracking, or anyone who needs an actual clinical diagnosis.

How it compares

Buoy Health sits in a different category from most wellness apps. Tools like Headspace focus on mental wellness and stress management rather than physical symptom triage, so they serve a complementary rather than competing role. For users primarily interested in women’s health tracking, Flo Health offers cycle and symptom logging with AI-driven insights, but it is scoped narrowly to reproductive health and does not attempt the broad symptom-to-care-path guidance that Buoy provides.

Within the symptom checker space, Buoy’s differentiator is the Harvard clinical partnership and the explicit care navigation step. General consumer health search tools surface information; Buoy attempts to surface a recommended action. Whether that distinction is meaningful in practice depends on how much weight a user places on the clinical sourcing behind the model.