Question 1

What exactly does the AI condition score mean?

Accepted Answer

The 0–100 score rates how attractive the lot is to an experienced salvage buyer given price, visible damage, title document, odometer, run-and-drive status, airbag deployment and known auction history. 70+ is a Good Deal, 45–69 is Average, below 45 is Risky. The 'explanation' field always tells you what drove the score so you can sanity-check it.

Question 2

Which auctions are supported — Copart, IAAI, others?

Accepted Answer

Copart and IAAI are fully supported — paste either a lot URL or a VIN and we fetch the listing photos, damage description, odometer, title document and auction history. Other US auctions (ADESA, Manheim) are on the roadmap.

Question 3

How accurate are the repair-cost estimates?

Accepted Answer

Estimates are given as a min–max USD range, broken down by part (bumper, fender, hood, suspension, etc.) and action (repair vs replace). They are pre-bid budgeting numbers based on what the AI can see in the listing photos — not a shop quote. Always get a real body-shop inspection before wiring money. Hidden frame or mechanical damage can shift the range significantly.

Question 4

Does the AI detect auction flip scams?

Accepted Answer

Yes. When a VIN has prior auction records and carpybara detects the re-listing pattern (car sold with severe damage, then re-listed by a non-insurance seller with a suspiciously clean exterior), the score is penalised and an explicit 'repair-flip' warning is added to the report.

Question 5

What about structural damage — can AI catch that from photos?

Accepted Answer

It can surface the tells that experienced buyers look for: asymmetric bumper gaps, crushed frame rails visible under the hood, wheel alignment issues, door gaps, airbag-deployment status, and salvage-title language. It cannot replace a frame machine or a structural scan. If the AI flags 'possible structural damage', treat it as a strong 'inspect in person' signal.

Question 6

Does it work on flood, burn or rollover cars?

Accepted Answer

Yes. The prompt is tuned to flag flood/water damage, burn/fire damage and rollover lots specifically — these are the highest-risk categories on Copart and IAAI, and the score is capped accordingly (typically below 45). The report also cross-checks the title document ('Cert of title — salvage flood', 'Certificate of destruction', etc.) with what's visible in the photos.

Question 7

Can I use this on European, Korean, Japanese or Chinese VINs?

Accepted Answer

Yes — the VIN decoder accepts any 17-character VIN from any manufacturer. The full AI damage analysis only runs on listings that currently exist on Copart or IAAI (since that's where the photos come from), but a VIN alone still gives you recalls, safety ratings and any historical auction data.

Question 8

What AI models power the inspection?

Accepted Answer

Scoring runs in parallel across top models from OpenAI, Gemini, and Groq. All models see the same set of listing photos; the final score is the one we're most confident in given the consistency of outputs. This avoids single-provider outages and hallucination on edge cases.

Question 9

Is the AI damage analysis free?

Accepted Answer

A free allowance of checks is available to registered users. Heavy users can upgrade to Pro (5 checks/month) or Ultra (10 checks/month), or buy top-up packs of 5, 30 or 100 one-off checks. Payments are securely processed by Stripe. See carpybara.ai/pricing.

Estimate repair risk before you bid

Estimate repair risk before you bid