In development

Point at the job. Get the next step.

AR Mechanic walks you through automotive procedures utilizing your phone's camera, labeling key points and giving you directions one step at a time.

Oil level · oil change · air filter, to start

oil filler cap engine oil dipstick Check engine oil level STEP 1 OF 5 Park on level ground, engine off, parking brake set. Let it cool first. clean rag gloves

How it works

From a camera frame to a finished job, without leaving the live view.

01

Scan

Point the camera at the part or area you're working on. AR Mechanic analyzes the frame and tells you what it sees.

02

Pick the job

Choose from the tasks it can guide for that part, like checking the oil level. Back out any time without rescanning.

03

Follow the step

Labels sit on the live view, a reference photo shows where to look, and a list of parts and tools says what to grab.

04

Verify

The camera checks the result against what finished work should look like, then moves you to the next step.

What it does

Built like a workshop tool: it shows what matters and stays out of the way.

Layered annotations

Every step carries up to three layers. Safety notes stay on screen. Pro tips expand when you ask for them.

  • safety
  • main
  • pro tip

Parts and tools, per step

Each step lists what it needs as chips, so nothing sends you back to the toolbox halfway through.

Checked against the camera

Each step has short visual cues the camera looks for before it advances.

  • cover removed
  • housing visible

Reference photo

A small reference shot with the target labeled sits over the live view. Tap it to enlarge.

Step back through the work

When the last step is verified, walk backward through completed steps to review what you did.

Session history

Finished jobs are saved on your phone with the task, the part, the time, a snapshot, and your notes.

Where your frames go

To the provider you choose, on your key. Nothing goes through Orbital AI.

AR Mechanic needs a vision model to understand what the camera sees. You bring the API key, so you pick who runs it. Gemini works today and has a free tier, and support for more providers is on the way. We run no server between your phone and your provider, and there is no account to create.

That means camera frames do leave your phone. They go straight to your provider under the terms of your own key.

  • Your key, in the keychainStored in the iOS Keychain or Android Keystore on your device.
  • What gets sentA compressed camera frame, to the provider your key belongs to, when the app analyzes a scene or verifies a step.
  • What Orbital AI seesNothing. No server of ours is in the path.
  • What stays on the phoneSession history, snapshots, and your notes.

Where it stands

A working proof of concept, built in the open.

The first goal is simple: show that a phone camera and a vision model can walk a first-timer through basic maintenance and know when each step is done.

There is no App Store or Play listing yet, and nothing here is a launch announcement.

  1. NOW
    Proving the loopChecking oil level, changing oil, and changing the air filter, with step verification.
  2. NEXT
    Specific engines, more providersGuidance tied to a particular engine, and keys from providers beyond Gemini.
  3. LATER
    Wider job listMore tasks as the first ones hold up in real garages.

Say when it's ready.

Leave your email and you'll hear from us the day there's something to try. Nothing else.