SnapKai: AI vision food logging without a mandatory subscription
A photo-based calorie and macro tracker that reads a meal photo with AI vision instead of asking people to search and log every ingredient by hand.
Visit SnapKaiProblem
Calorie and macro tracking apps typically ask people to search a food database and log every ingredient manually, a habit most people abandon within days. SnapKai needed a faster path from photo to nutrition data, priced without forcing a subscription on people who only want to log occasionally.
Architecture
Frontend
A React/Redux progressive web app, with a separate Astro marketing site.
API
Two .NET 10 APIs backed by Postgres.
Database
PostgreSQL.
Infrastructure
AWS and Docker on EC2, with TLS and Google/Microsoft OAuth in production.
AI components
AWS Bedrock vision models that read a meal photo or a printed nutrition label and return structured calorie and macro data.
Engineering decisions
- The credit model, one-off top-ups rather than a mandatory subscription, was the central product decision, and it became the base other Corvidae products later inherited.
- Barcode, manual and repeat/favourite logging were kept always free, so the AI vision scan is the one paid action rather than a gate on basic use.
- The product shipped as an installable progressive web app with Android Trusted Web Activity scaffolding, rather than committing to two separate native codebases from the start.
Challenges
- Getting AI vision accuracy on real, imperfectly lit meal photos and printed nutrition labels to a level worth paying for.
- Running OAuth, Stripe billing, analytics and ad conversion tracking correctly through a PWA rather than a native app shell.
- Keeping credit-based pricing (a photo scan costs more credits than a label scan) simple enough to understand at a glance.
Outcome
- Launched and live at snapkai.com, with TLS, OAuth, Stripe billing and analytics verified working in production.
- An installable PWA with a daily food log, weight tracking, reports and a referral credits program.
- Became the codebase Lernoa was later forked from.
Services demonstrated
Other case studies
Want something built this way?
Tell us what you want to build. We reply within one business day with an honest read on scope, cost and timeline.
Start the conversation