Software that turns abandoned bikes into graded, priced, carbon-verified inventory.
RescueOS is the technology behind Rescue Bikes. Computer-vision grading, recovery logistics, digital bike passports and verified CO2 certificates, so any city or workshop in Europe can plug in and put rescued bikes back on the road at scale. The Tallinn workshop is our pilot; the software is the product.
See an intake graded in seconds
Pick a sample bike and run the assessment. This prototype runs a transparent rules engine in your browser - the production system swaps in a trained ML model as data grows.
Pick an intake bike
Choose a sample bike, then run the assessment.
Run the assessment to see the grade, repair plan, price and carbon certificate.
Every intake becomes structured, machine-readable data - grade, repair plan, price and a carbon certificate - in seconds, with no expert required. This is the core IP that makes refurbishment repeatable and scalable.
Real computer vision (TensorFlow.js), running in your browser — detects the bike and grades it.
One pipeline, from waste stream to road
Cities and donors feed in bikes, the AI grades and routes each one, workshops refurbish against a generated work order, and buyers get graded, warrantied bikes.
For cities & associations
Report abandoned bikes, get optimised pickups, and receive automated CO2 and diversion reports for every batch.
For workshops
Receive AI work orders plus a built-in storefront, and get paid directly through the platform.
For buyers
Every bike ships with a digital passport: grade, repair history, warranty and verified CO2 saved.
Capture, detect, grade, price, certify
Capture
Intake photos and basic details are uploaded from any phone.
Detect
The vision model segments the frame and key components.
Grade
Each component is scored to produce an overall condition grade.
Price
Repair cost, resale price and margin are estimated instantly.
Certify
A digital passport and verified CO2 certificate are minted.
Under the hood, a computer-vision classifier - transfer-learned from open vision backbones and fine-tuned on labelled bike-intake images - feeds a pricing model. Today a transparent rules engine bootstraps the outputs and labels training data; it is replaced by the ML model as volume grows. That is the data flywheel.
A software company, not a bike shop.
The hands-on work stays local; the engine that makes it repeatable is software.
Repeatable
A new city is a software onboarding. Local workshops do the hands-on repair work.
Globally scalable
The same engine runs in Riga, Helsinki, Berlin and Amsterdam without rebuilding.
Innovative
No one else applies CV grading plus automated carbon certification to the abandoned-bike waste stream.
Built by Johannes Gontrum, Co-Founder & CTO (ML/NLP engineer, MSc Uppsala University, ex-Retresco, spaCy contributor). RescueOS is at working-prototype stage; the trained vision model, dashboards and CO2-verification methodology (with TalTech) are in active development.