Real computer vision · not a mockup

Upload a bike photo. Watch the AI grade it.

This runs a real pretrained vision model in your browser (TensorFlow.js). It detects the bicycle in your photo, draws the detection box, then a transparent grading engine estimates condition, price, repair cost and CO2 saved. Try it with any bike photo from your phone or the web.

1. Give it a bike photo

📷 Click to upload or drop a bike photo here

JPG / PNG · stays on your device

Or a sample:

Your photo and the AI's detection boxes will appear here.

Loading AI model…

Initialising…

2. AI assessment

Run an assessment to see the grade, component analysis, pricing and a verified CO2 certificate.

B

Standard refurbishment

Refurbished channel

Detected: bicycle (—%)

€—

Resale price

€—

Repair cost

—%

Margin

— kg

CO2 saved

Verified impact certificate

— kg CO2

Digital passport

RB-TLN-000000

How this works (honest version): the bicycle detection above is real AI inference on your photo. The condition estimate is a v0 heuristic from image features (detection confidence, sharpness, exposure, colour/rust signal). v1 replaces it with a model fine-tuned on our proprietary dataset of graded bike intakes.

For workshops · SaaS

Work orders, generated automatically

Every graded bike becomes a costed work order. Mechanics stop guessing and start fixing. This is the screen a partner workshop sees.

37

Bikes in queue

€2,140

Projected resale

4.6h

Avg refurb time

67%

Avg margin

Passport	Grade	Recommended work	Parts	Resale	Status
RB-TLN-004811	A	Service drivetrain, true wheels	€14	€182	In progress
RB-TLN-004812	B	Replace brake pads, chain	€21	€131	Ready
RB-TLN-004813	C	Full service, new tyres	€33	€74	Queued
RB-TLN-004814	D	Harvest: chain, derailleur, wheel	—	€0	Parts
RB-TLN-004815	A	Cosmetic only, ready to list	€6	€205	Ready

For cities & associations · SaaS

Abandoned-bike reporting, on autopilot

Cities feed in abandoned bikes; RescueOS inventories them, routes them to workshops, and generates the diversion and CO2 reports a municipality needs. This is the city's dashboard.

1,284

Bikes recovered (YTD)

71%

Diverted from scrap

321 t

CO2 avoided

€18k

Disposal cost saved

Recovered bikes by month

Jan

Feb

Mar

Apr

May

Jun

Auto-generated; exportable as a council ESG report.

Outcome breakdown

Refurbished & rehomed612 bikes

Free-tier (community)298 bikes

Parts harvested261 bikes

Awaiting assessment113 bikes

Each outcome carries a verified CO2 figure rolled up into the totals above.

For developers · the product we sell

One API call grades a bike

RescueOS is sold as an API. Refurbishers, marketplaces and cities send photos and get back structured, machine-readable inventory. Below is the live contract — the "Try it" button runs the exact same in-browser model the Grading tab uses.

POST /v1/grade

curl -X POST https://rescueos-api.rafi-drifly.workers.dev/v1/grade \
  -H "Content-Type: application/json" \
  -d '{
    "images": ["https://example.com/bike.jpg"],
    "bike_type": "city",
    "region": "EE"
  }'

Live now at rescueos-api.rafi-drifly.workers.dev — an open reference endpoint, no key needed. /v1/feedback takes the realised outcome and becomes a new training label — the data flywheel. Production adds auth + per-call metered billing.

Response

{
  "status": "ready",
  "hint": "Grade a bike on the AI Grading tab,
           then press Try it to see its JSON here."
}

A reference server implementation (FastAPI + Cloudflare Worker) ships alongside this app — and it's live now at https://rescueos-api.rafi-drifly.workers.dev. See rescueos-api/README.md.