Salamati exists because every existing health app puts friction between the moment something happens and the moment it's logged. We removed the friction.
Most people who try health tracking apps quit within two weeks. Not because the idea is wrong — tracking your nutrition, sleep, and workouts genuinely helps you understand yourself. They quit because logging is exhausting.
Open the app, search a database of 8 million foods, find the one that's close enough, estimate the portion size, log it, repeat for every component of the meal. By the time you're done, you've spent three minutes on a meal that took thirty seconds to eat. Do that three times a day, every day, and your brain starts subconsciously avoiding meals that are hard to log. That's not tracking anymore — that's optimizing your life around the app.
Fitness is the same: log exercise, pick a category, enter sets and reps, estimate burn. Supplements: tap through your stack every morning. Sleep: fill in a form.
The apps are technically capable. The problem is the input model. Tapping, searching, and form-filling is the wrong interface for the real-time, often-while-doing-something-else moments when health data gets created.
Talking is how people narrate their lives. When something happens, you describe it. You say it out loud, you text it to someone, you tell your partner when you get home. The moment is already being verbalized — we just weren't capturing it.
Voice is fast (under 12 seconds per entry, on average). It's natural — you don't have to remember any syntax. It works while you're doing other things. It captures nuance that a form can't: "I had a rough workout, felt tired but pushed through" is a different health signal than just "45 min strength session," and Skrypt captures both the structured data and the context.
The reason voice didn't work in previous apps was the technology. Transcription was slow and error-prone, and converting natural language into structured health data required rigid templates. Both of those constraints dissolved in the last two years. Deepgram gives us sub-second transcription with clinical accuracy. Claude gives us AI classification that understands how people actually describe food, movement, and mood — not just exact food names and exercise categories.
Salamati is a voice-first PWA (Progressive Web App) that works on any device with a browser. The technical stack is intentionally minimal — we pick the best tool for each specific job and don't add complexity without reason.
Real-time voice transcription. Sub-second latency, high accuracy across accents and ambient noise. Your voice audio is transcribed and immediately discarded — never stored.
AI classification and macro estimation. Understands how people actually describe food ("the bowl thing from that place"), movement, and mood — not just database lookups.
Encrypted health data storage with row-level security. Your data is isolated per user and can be exported or deleted at any time — no lock-in.
The result is a system that's genuinely fast. From "tap the mic" to "structured health entry stored" takes under three seconds on a typical connection. That speed matters because even a 10-second logging experience trains you to avoid it.
I built Salamati because I kept failing at health tracking — not from lack of motivation, but from the friction. I tried MyFitnessPal, Cronometer, Whoop, and half a dozen others. They all worked for a few weeks and then stopped working when real life got busy.
The insight I kept coming back to: the apps weren't wrong about what data mattered. They were wrong about how to collect it. I'd describe my meal to my partner at the dinner table in five words — why did the app need twenty taps?
I'm also building Skrypt Health, an AI company focused on healthcare operations. The consumer health product and the clinical product inform each other — understanding how patients narrate their health day-to-day makes us better at building tools for clinicians, and vice versa.
Salamati is a product of Skrypt Health (Invite Technologies Inc.), an AI company based in Toronto, Ontario. We build two products:
Consumer voice-first health tracking. Nutrition, fitness, supplements, sleep, mood — one AI companion.
AI front office for dental and medical clinics. Appointment booking, intake, and patient communication — automated.
Both products are built around the same conviction: the right interface for capturing health information is natural language, not forms. The consumer product and the clinical product share infrastructure and research, which makes both of them faster and better.
Your voice audio is transcribed and immediately discarded. We never store raw voice recordings. Your health data is never sold to third parties, never used to train AI models without explicit opt-in, and you can export or delete everything at any time. Read the full privacy policy →
Web app is live now. iOS native app is on the waitlist. Free during early access — no credit card.
