What the AI looks at
- The photo (object recognition, brand and model when readable).
- Any text you typed before triggering the suggestion.
- Your destination's purchasing-power index (a sofa in Ottawa is valued differently than the same sofa in Tunis).
- General market data for the category, with a bias toward replacement cost in Canada.
Where it does well
- Common household categories: furniture, kitchenware, kids' clothing, electronics.
- Items with a visible brand and model (laptops, appliances, power tools).
- Photos taken at decent light with the object centered.
Where to double-check
- Jewellery, watches, art, antiques — use a formal appraisal. The AI will guess, but the variance can be very large.
- Custom or hand-made items — the AI compares to mass-market analogues.
- Sentimental items with no resale market — the AI may return $0 or a placeholder. Set the value yourself.
Reporting bad suggestions
If a suggestion is wildly off, tap Feedback on the item screen. We use the reports to refine the prompts. The Audit-prices-with-AI helper on the Import screen does the reverse pass: it flags any item where the value looks inconsistent with the description.