AI part identification
Trained specifically on automotive headlights. It tells apart years, generations, features, and left or right variants.
List a headlight in a few clicks
Lotwize identifies the exact headlight and fills the title and item specifics from your customizable template.
What it does
Everything between the photo and the listing.
Photo handling
Pictures with clear text (part numbers, stickers) are turned the right way up if needed and moved to the end of the listing, before your stock number photo if you have one. Drag and crop photos to your liking.
Personalized listing
It’s a template, and it’s yours to edit. Adjust the titling and item specifics to your style, and bring your own HTML description template if you already have one.
Automatic condition placement
Describe the condition briefly in a text box and it will automatically be placed wherever appropriate within your description.
Comparing similar listings
See what comparable headlights are listed for right now, narrowed to your exact part, with components and mismatched trims dropped, and ranked against the condition you typed.
The Lotwize Autolight Identification Model Benchmark
A model tuned specifically on 300,000 headlight images, measured against an anchor set of parts verified by hand.
1,611parts in the anchor set, each verified by hand
| Make | 98.9%n=1,611 | |
|---|---|---|
| Model | 97.3%n=1,611 | |
| Year range Overlapping the true range | 99.3%n=1,611 | |
| Side | 98.9%n=1,611 | |
| Bulb type Parts whose bulb cannot be read from a photo are not counted. | 99.7%n=1,433 | |
| Optic To limit redundant optic naming, it is only asserted on generations that ship different optics across the same bulb with exceptions for the general titling aesthetic. | 99.7%n=1,522 |
Zero false positives across 792 feature records and all 22 tokens, on 1,611 parts verified by hand.
Across 1,611 parts verified by hand in our anchor set, scored over 792 feature records and all 22 tokens, the benchmark delivered zero false positives. Having to stop and correct the model’s output is inconvenient, and in some cases expensive.
Where a label or a feature is one the model is known to struggle with (across multiple evaluations of the model), it is not allowed to assert it. Those are held behind a filter gate and reach you as a checkbox to tick instead, so a field we cannot stand behind is presented to you rather than asserted.
Measured on 1,611 parts verified by hand across 37 makes, from our anchor set of real listing photos. Your results depend on photo quality, orientation, and how much of your vehicle’s generation we have seen. Our zero false positive claim and the benchmark behind it are based on our anchor photos, so your results may or may not reflect what we have recorded.
Figures from the current board, 7 September 2026.
Contribute to our anchor set
Help improve the model by submitting corrections.
Correct an identification and submit it to our anchor set. Once it passes manual review, we target these gaps within the model to improve overall accuracy.
Anchor Submitted
Questions
FAQ
Does it list to eBay for me, or do I still do that part?
It lists for you. Lotwize connects to your eBay account under your consent, and publishing happens entirely inside the app. The listing goes live with your set business policies and description, with condition inputted where necessary.
What does it cost, and where does it work?
Lotwize is a monthly plan. Pick the size that matches how much you list: $49 a month for up to 100 headlights, up to $749 for 2,500. After that each headlight comes out of your plan, and anything past the allowance costs five cents a headlight more than your plan’s own rate.
Where do the prices come from?
You set the price. Lotwize does not suggest one. Every other step publishes as described.
What happens when it identifies the part wrong?
You correct it, and the correction is saved for review. The corrected part is also added to the anchor set (the confirmed examples the model is measured against). Every field the AI fills stays editable. Identifications are assistance; please review them before publishing.
Is it only for headlights?
Yes, and deliberately. The Lotwize identification model is trained purely on automotive headlights.