Gymeat: a workout tracker that counts what actually builds muscle.

The problem

Popular workout loggers count "sets of an exercise". That number lies: a bench press trains the chest directly and the triceps indirectly, an easy warm-up set counts the same as a set taken close to failure, and one-arm work is often counted twice. Lifters end up with totals that look right and training that isn't.

What we built

Gymeat turns every logged set into what it does for each muscle, and grades the week against published research.

• Fractional sets per muscle: 1.0 for a muscle trained directly, 0.5 for one trained indirectly, the counting method that performed best in Pelland et al. (2025).
• Only hard sets count. Sets more than four reps short of failure are left out of the totals.
• Unilateral work counts once per pair of sides, not twice.
• Weekly volume grades against published efficiency tiers and volume landmarks. Where no primary source exists, the app says so instead of inventing a number.
• In-session warnings when a muscle passes roughly eleven fractional sets in one workout, with a suggestion to move volume to another day.
• Every number links to its source, inside the app.

How it's built

A Cloudflare Worker API in TypeScript (Hono, OpenAPI) over PostgreSQL through Hyperdrive, and an Expo / React Native client for Android and the web. Exercise media is served from Cloudflare R2 and the web app is deployed to Cloudflare Pages.

What clients get from it

Gymeat is a product where the hard part is the domain, not the CRUD: turning research papers into precise, testable rules and explaining them clearly to users. If your product depends on expert knowledge such as health, finance, logistics or compliance, that's the kind of work we enjoy. See our custom software development service.