Real numbers, not vibes.
Connect once, with your account's permission, and your assistant can see what you actually lift — then act on it.
- Reads your history — sets, PRs, estimated 1RMs, muscle-volume balance, consistency, week-by-week trends.
- Writes real programming — routines and weekly plans it drafts appear in the app, scheduled and ready to run.
- Logs for you — dictate a set or a run to your assistant; it lands in your training log.
- Judges sessions like the app does — the same honest verdict engine, available to your agent.
- Estimates starting weights — for lifts you've never logged, inferred from the ones you have.
How did this training block actually go?
Ten weeks, 48 sessions, one missed day. Squat top set moved 245 → 290 lb; deadlift e1RM is 413 lb, a PR. Your shoulder press has been flat for six sessions — that's the one to change.
Fix it in next week's plan.
I've updated Push Day: shoulder press drops to 105 lb for a 3×8 rebuild, with lateral raises moved up. It's on your Monday.
The tool surface.
Every tool is scoped to your account and gated by sign-in. This is the actual production surface, live today.
Training history
Workouts with full set detail, per-exercise history, PRs, cardio, body metrics, and a dense context snapshot built for agents.
The same engines as the app
Strength standards, muscle-volume vs. evidence bands, consistency streaks, stall detection, session verdicts, week in review.
Programming that lands
Create and revise routines and weekly plans; log sets and cardio with idempotency keys so a retry never double-logs.
Gated, audited, yours.
Agent access rides the same OAuth sign-in as the app — no API keys to paste, nothing shared by default. Disconnect anytime. Your data is never used to train models, never sold, never shown to another user's agent.
Meet your new
training amigo.
Coming soon to the App Store. Leave your email and you'll hear about it first — nothing else, ever.