Still searching

My bike was stolen, so I built an AI to look for it

I'd had her for 10 years. At 3am on 1 October 2026, someone took her from outside my flat in London. I filed a police report and heard nothing back, so I'm trying to get her back myself. A script on my Mac now checks every new bike listing on Gumtree and Facebook Marketplace and scores how much each one looks like mine.

Fred's stolen bike locked to a rack: a blue single-speed with a white rear triangle, red and yellow stripes and a white saddle
This was my bike. She was a beauty.
Listings checked
156,029
New in the last 24 hours
4,419
Possible matches (score 40+)
833
Highest score so far
74 / 100

Figures as of .

What happened

My doorbell camera caught the whole thing. The heist took seven minutes, from 2:56am to 3:03am. A man walks past, spots the bike, has a look around the building, takes it and leaves with a friend.

Doorbell camera still at 3am on 1 October: a man in a blue jumper wheeling the blue and white bike away
Six CCTV stills of the theft between 2:56am and 3:03am, labelled: sees bike, wants bike, looks for bike, takes bike, leaves with friend. Faces are pixelated.
The heist, 2:56am to 3:03am. Faces pixelated.

I filed a police report. Sadly, no response. So I needed a plan B.

A lot of stolen bikes get sold on Facebook Marketplace and Gumtree. The problem is volume: thousands of bikes go up on those two sites every single day, and I can't sort through them all. But a computer can.

Plan B: hello, Claude

The idea was simple. Search for bikes on these marketplaces. Log every new listing. Compare every photo to my stolen bike, give each one a score out of 100, save the results, and do it all again, every day.

I built the checker with Claude. It runs on my laptop, and so does the AI that looks at the photos, so there's no cloud service and no bill per image.

1

It reads the newest listings

A few times a day it opens Facebook Marketplace and Gumtree, newest first, and reads until it reaches listings it has already seen. It also searches the whole UK for words like "fixie" and "single speed", in case the bike has left London.

2

It scores every photo out of 100

Each thumbnail goes through CLIP, an AI model that matches pictures to descriptions. It's asked a checklist of questions about my bike, one at a time, plus how similar the photo is to my own photos of it.

3

It sends the best ones to my phone

After each run, the top 20 new listings arrive on my phone as a grid of photos with their scores. If one scores 40 or more, the notification is marked urgent. A dashboard on my laptop shows every listing it's ever seen.

The 'Operation: Find My Bike' dashboard: a green-on-black control room showing a queue of bike listings, the one being scored, its feature scores and running totals
A few prompts later, the dashboard. It's a bit over the top. 45,610 listings checked by 9 October.

The questions it asks about every bike

Colour · counts double

Is it blue and white, or black, red, silver, green…?

Type

A road or single-speed bike with thin tyres? Or a mountain bike, BMX, e-bike, kids' bike, or not a bike at all?

Saddle

White, black or brown?

Stripes

Coloured stripes on the frame, or plain paint?

Asking one question at a time works much better than giving the AI one long description, because it struggles to tell which colour belongs to which part of the frame. The answers are combined so that a bike has to match on everything to score highly. A blue mountain bike won't get far.

To set the bar, I scored my own photos of the bike against each other. They came out between 53 and 65. In early tests, the best real listing scored 44. So anything over 40 gets a look from me: a false alarm only costs a glance. It's a ranking, not a probability. So far nothing that scored high has turned out to be mine.

It's also deliberately polite. It waits between page loads like a person would, caps how much it reads each day and backs off for a day or more if a site pushes back. Gumtree blocked it on day one, so now it reads one Gumtree page every 20 minutes.

The stolen bike from the side: blue top tube and forks, white rear triangle, red and yellow stripes, white saddle

Have you seen this bike?

Blue and white single-speed, red and yellow stripes

  • Blue top tube and front forks
  • White rear triangle and down tube ends
  • Red and yellow stripes where the colours meet
  • White saddle, flat handlebars
  • Stolen in London at 3am on 1 October 2026
Tell me where →

Updates

Newest first. If the bike turns up, it'll be at the top.

  1. Stats

    156,029 listings checked, and still looking

    Just over five weeks in, the checker has looked at 156,029 listings. It finds around 4,400 new ones a day. 833 have scored 40 or more, which is enough for me to take a look. The highest score so far is 74.

    The bike hasn't turned up yet. The checker keeps running.

  2. LinkedIn

    151,992 people saw the post. Nobody's seen the bike

    My LinkedIn post about the theft passed 151,992 impressions, with 693 reactions and 54 comments. A great way to get lots of LinkedIn impressions. A terrible way to lose a bike.

  3. Stats

    45,610 listings checked, 221 close matches

    My bike next to a grid of nine blue road bikes from marketplace listings that scored as close matches

    One week in. The checker had looked at 45,610 bike listings and flagged 221 as close matches. Lots of blue bikes. None of them mine.

    I posted the story on LinkedIn the same day.

  4. Build

    The first run

    The checker's first full run read 2,011 Facebook Marketplace listings in one go. Gumtree blocked it within a day, which is how it ended up with its very polite one-page-every-20-minutes routine.

  5. The heist

    Stolen at 3am

    Between 2:56am and 3:03am, my bike was taken from outside my flat in London. The doorbell camera saw everything. I filed a police report and never heard back.

To be continued…

We haven't found her yet. Sorry, I wish this post had a happy ending for you.

But if she ever gets posted online, the checker will find her. So it's game on.

I still believe!

© 2026 Cluckin About Ltd.

Hatched in London