JSON the model can read.
Full schema. UUIDs everywhere. Soft deletes filtered out. Checksum at the top. The same shape docs/SCHEMA.md publishes for developers — no mystery fields, nothing renamed for "branding."
For the coach
Hand a clean JSON to anyone.
You write the program. The lifter trains. The data should round-trip without your spreadsheet ever needing a manual cleanup pass. flexRep ships a readable JSON with full schema, a CSV in two formats, and reversible importers — so handoffs in either direction are five seconds, not five hours.
Where flexRep meets you.
You probably manage your athletes in a spreadsheet. flexRep doesn't replace that — it feeds it. A coach gets a portable, machine-readable export and a per-set provenance trail. Imports from other apps stamp a batch ID so a bad import is one-tap reversible.
AI coach handoff, already drafted.
Drop the JSON into your model of choice with our suggested prompt and get a one-page markdown summary: training frequency, exercise selection, working rep ranges, average RPE per lift, recent PRs, concerning patterns. The lifter's data goes to the LLM you trust, not ours.
Imports that can't poison your data.
Every CSV import — Strong, Hevy, generic — stamps a shared batch ID on every row. Reverting removes only that batch; native logs are left alone. Imports are a data-poisoning risk by default; the batch boundary defangs them.
One JSON. One prompt. One markdown summary.
A typical coach hand-off: the athlete shares the export, the coach drops it into their LLM of choice with the suggested prompt, and forty seconds later there's a markdown summary on every lift, every block, every concerning pattern. The lifter's data goes to the model you both trust — never through a flexRep server, because there isn't one.
Last 12 weeks · J. Stenmark
Frequency. 4.1 sessions per week · 47 total. Two missed weeks in early March (reported flu).
Lift selection. Bench, squat, deadlift, OHP as primaries. Six accessory rotations across the block.
Working rep ranges. 3–6 on primaries (62% of working sets), 8–12 on accessories (33%), 12+ on isolation finishers (5%).
Average RPE. Bench 7.8 · Squat 7.6 · Deadlift 7.9 · OHP 8.1.
Recent PRs. Bench e1RM 285 → 295 (+10 over 6 wks). Squat e1RM 395 → 415 (+20 over 8 wks).
Concerning pattern. Squat RPE in the 80–90% zone has drifted +0.5 over four weeks at constant relative load. Recommend a deload week or back-off.
- 01 Athlete exports JSON
Settings → Export → JSON. One tap. Share-sheet to email, AirDrop, or message.
- 02 Coach pastes into their LLM
Claude, ChatGPT, Gemini, or local. Use the "Coach handoff" prompt from /data.
- 03 Markdown summary lands
One page, paste into your coach notes. Forty seconds, end to end.
Five features doing real work for you.
Out of the The Coach's thirty-six daily features, these are the ones that earn their seat at the rack.
Coach Digest — single-tap markdown recap
Athlete name · week range · federation / goal · volume by muscle group · top sets per top-5 lifts · per-lift stall verdict · routine adherence vs prescription · readiness signals · top three set notes. One tap. Paste into your coach notes. Honors privacy mode (redacts exercise names to muscle groups).
See itVerdict-mode AI with typed action items
On-device. Returns a typed action — add_set · drop_volume · deload · see_clinician · stay_course — with a target, justification, and confidence score. Low-confidence verdicts (< 0.5) get explicit "Low confidence" framing instead of false certainty.
See itFull backup bundle with checksum
Everything in one verifiable file: programs, mesocycles, microcycles, prescriptions, body photos, personal records, meet-day plans, routine revision history. Round-trip tested every release.
See itReversible imports, every source
Strong, Hevy, generic CSV, HealthKit. Each import stamps a batch ID. One-tap revert removes only that batch — even your demo data ships behind a dedicated "Remove demo data" toggle.
See itPrograms you can write and share
Program → Mesocycle → Microcycle → Routine — a real graph, not a string. Export the graph; another coach can read it. Routine revisions back up with the rest of the bundle.
See itHow it stacks up for a coach
Coach-friendly apps tend to lock data behind a paid tier or a proprietary "coaches dashboard." flexRep's position is that your athlete's data is theirs, and a clean export is the contract that makes coaching possible.
Competitor
flexRep wins on
flexRep loses on
TrainHeroic
No paywall on the athlete's side · full JSON · 25-entity schema
Built-in roster + comms surface
TrueCoach
Athlete owns the data · LLM-ready JSON · methodology you can verify
Coach-side dashboard for managing N athletes
Strong
Reversible imports · two-format CSV · checksum bundle · programs as a graph
Brand recognition with athletes
Spreadsheet only
A model that survives a phone wipe, an app shutdown, or you
The freedom of an empty cell
Refusals are also features.
- We don't paywall export.
- We don't sell anonymized athlete data.
- We don't silently overwrite imported sets.
- We don't require a coach account.
A few other lifters who feel at home here.
The Powerlifter
Numbers that hold under load.
The Bodybuilder
Volume the literature would recognize.
The Program Runner
Your block, as a real graph.
The Lifter Managing Pain
Train hard. Train smart. Train forever.
The Returning Lifter
Show up. Tap. Lift.
The Climber
Hangboard edges and V-scale grades, first-class.
The Olympic Weightlifter
Sinclair, not Wilks — a separate scoring discipline.
The Strongman Athlete
Yokes, sleds, carries, atlas stones — modeled.
The Grip Athlete
Joint angles, per-set grips, fine forearm radar.
The Hybrid Athlete
Strength + conditioning, one logger, honest math.
Hand the JSON to the model you trust.
All of this, on your device. In your data. No account required to start.