Learns from every run. Paces every climb.

Grade turns the runs you’re already logging in Garmin, Intervals.icu, or TrainingPeaks into a real model of how you climb, descend, hold up over hours, and even how you respond to heat. Whether it’s a vertical K or a hundred-mile ultra, that same model plans your pace, fueling, and effort for the actual course, and keeps adjusting while you run it.

Works with

The problem

A grade-adjusted formula isn’t the same as a model of you.

Elevation-aware pacing tools, including the one already built into your watch, apply one published cost curve to every runner: the same penalty for a 9% climb whether you power-hike it efficiently or grind it flat-footed. Grade throws that shared curve out. It fits your own curve from your actual climb and descent splits, so two runners on the same course get different plans, because they’re different runners.

Training model

Built from the runs other platforms throw away

Most platforms need a number you type in, or a maximal test, to set your zones. Manual numbers go stale fast, and max-effort tests are rare, they don’t say much about how you’re doing this week. Meanwhile most of your training, the easy runs and long climbs sitting below threshold, gets treated as noise.

Grade uses that history instead. It fits your thresholds directly from how your heart rate responds to pace and power across everyday runs, tuned for the terrain you ran them on. Near-max efforts help when you have them, but they’re not required. Recovery jogs, run-walk long days, even hikes with the poles out, all count as data.

Thresholds Updated 2 days ago
LT1 heart rate 158bpm ±3
LT2 heart rate 171bpm ±4
Critical power 318W ±9

Fit from 46 runs over 10 weeks

Terrain

Modeled everywhere, not patched on afterward

That climb-and-descent model doesn’t stop at pace. The same profile carries through vertical speed, surface underfoot, and heat load, from sessions you’ve actually run in it. A treadmill session at 12% incline and a technical descent on singletrack both feed the same athlete profile, kept in context.

Sable Ridge climb
Grade-adjusted pace 4:52/km
Raw pace 6:14/km
Climb +9% pace +1:22/km HR +11bpm
Descent −6% pace −0:18/km HR +6bpm

Live control

A treadmill that runs the workout for you

Pair a compatible Bluetooth treadmill and Grade drives the speed and incline through a structured session for you, including built-in tests like an LT1 heart-rate drift protocol at a fixed grade. Your watch keeps recording as usual, Grade just runs the workout.

It’s the same model that estimates your thresholds from training history, just running live. Your targets come from what your body is doing that day, not a plan you typed in a month ago.

Treadmill · Live 18:24 / 32:00
Pace 4:14/km
Incline 11%
Heart rate 164bpm
Cadence 168spm
Power 312W
Interval 4 of 6 · LT1 drift test

Next: 2:00 recovery at 9% incline, then interval 5 at 15%

Provenance

See why a number changed, not just that it did

Most platforms either publish a formula and leave the inputs to you, or estimate everything behind a sealed model you can’t inspect. Grade keeps a record for every estimate: which runs it came from, how much weight each one carried, and the rule that produced it. When a threshold moves, you can see the runs that moved it. When Grade isn’t confident, it says so instead of guessing quietly.

12 runs LT1 · 158 bpm+6 vs. 3 wks ago High confidence

Driven by

Oct 12 · Long Run Oct 28 · Tempo Nov 3 · Hike, Poles Out

Integration

It runs alongside the tools you already trust

Grade isn’t another place to log your runs. It reads history from Garmin, Intervals.icu, and TrainingPeaks, learns from it, and hands back thresholds, baselines, and race plans you can act on or sync back to your calendar. Outdoor trail runs and indoor treadmill sessions feed the same athlete model; your watch stays the recording authority.

On race day

The same model, live on the course.

Load a race’s GPX and Grade turns your training model into a plan for the climbs, the descents, and the long middle, not a flat-pace calculator. Once you’re moving, it watches your effort and re-plans what’s ahead: what you planned (Baseline), what you should do right now (Adjusted), and what you’re actually doing (You). As you fatigue and conditions change, like heat, Grade adjusts the plan instead of leaving you to guess.

Before the start, Grade pulls the race-day forecast, checks it against how you’ve handled heat in training, and hands you a fueling plan, carbs, sodium, fluids, timed to the course, plus a finish-time range instead of a single flat-pace guess.

The same numbers push straight to a Garmin data field, so the plan is on your wrist even when your phone stays in the drop bag.

Structural differentiation

How Grade differs

Other platforms Grade
Race-day pacing Flat finish-time predictors, split-average pace Point-by-point plan from your GPX that adjusts as the race goes
Fueling & finish estimate Generic race guides, one flat-pace finish time Carbs, sodium, and fluid plan from forecast and heat training, plus a finish-time range
Terrain Flat-equivalent or simple correction Modeled throughout: GAP, climb/descent efficiency, heat
Threshold inputs User-entered FTP, LTHR, zones Derived from submaximal load↔heart-rate curves
Model visibility Formulas published, or a sealed black box Evidence stream with per-estimate provenance
Compute timing After upload During the activity: treadmill control, live protocols, race replan

The result

One model. Every run makes it sharper.

Thresholds, baselines, and pacing targets update as you train, not on a schedule you manage yourself. On a compatible treadmill, the same model runs your workout live. On race day, it paces your climbs and adjusts when the day doesn’t go to plan. Grade doesn’t replace your watch or your training log, it reads what they already know and puts it to work.

Get in touch

Let’s talk.

Early access, partnerships, or questions. Invite-only for now.

Email us