Vehicle range predictions & route planning

Machine-learning route forecasting — see how much range a drive will cost you before you set out.

See it work

Plan a trip and the model reads the road ahead — distance, elevation, the shape of the drive — then forecasts what it'll cost. It learns from your own past recorded drives, so the numbers are tuned to how you actually drive. Here: San Francisco Ferry Building → Oakland City Hall comes back −7.59 kWh — starting at 72.6% and landing at 66.7%.

Request access

In private beta — invite-only, no cost. Tell me a little about your use (EV range, or another signal you'd want predicted along a route) and I'll get you onto the whitelist.

Your details are only used to set up access — never shared or made public. Site usage stats are anonymous.

Already whitelisted? open the app ↗

What you're signing up for

Access is granted by hand while the beta is small — that keeps it stable and lets me hear what early users actually need. There's no cost during the beta, and no obligation once you're in.

To be upfront: the beta runs on open, free services today, which keeps it free — but also caps how many people it can serve at once, so access grows gradually. A fully hosted version with room for everyone would carry real running costs, so expect a paid option down the line. Early beta users will always be dealt with fairly.

What it does

Plan the route

Lay out a trip and the app reads the road ahead — distance, elevation, the shape of the drive from real GPS data.

Forecast a signal along it

A machine-learning model predicts a quantity that changes as you travel. Today that's electric-vehicle battery percentage — how much charge a given route will actually cost you.

Works without a signal

Drives run through canyons, tunnels, and dead zones. The app keeps recording and working offline, so a dropped connection doesn't cost you the trip.

Built to generalize

The model isn't tied to batteries. Give it any target you can measure along a path and it learns to predict that instead — range is just the first thing worth knowing before you go.

And it's accurate

These are real numbers from my own logged drives — the app's own Model Health page. Across 28 drives, arrival battery % lands within 0.85 points on average, and the error stays flat as trips get longer — nailing a long route is the hard part, so a flat trend is the win.

Model Health stats: 28 drives logged, mean arrival-SoC error 0.85 percentage points
28 drives · 0.85 pp mean arrival error
Charts: arrival-SoC error over time hugging zero, and accuracy staying flat with trip distance
Error stays near zero — and flat with distance

Vehicle support

No car to connect? Try demo mode. It drops you into the full experience with a sample vehicle — plan routes and watch the live forecasts — so you can see how it works before you ever link a real EV.

Today the forecasts run on electric vehicles, and Rivian works out of the box — connect your account and the route range comes back straight away.

Other EV makers — Tesla, Hyundai, Kia, VW, and beyond — need a tester before they're supported. Each manufacturer talks to its vehicles differently — a different account connection and API for reading the car's live state — so I pair with an owner of that make to wire it up and confirm the numbers on real drives. If you drive one and want yours connected, say so in the access request above — I'll bring you in as a tester for your make.