How It Works
How Kitar grades your devices.
Upload intake photos. Get a grade and a live-demand price quote in under 10 seconds. No manual inspection sheet, no price-list spreadsheet, no back-and-forth.
Four steps
From photo to price in one workflow
The process runs entirely inside Kitar. Your staff never leave the interface to check another tool.
Capture intake photos
Staff photograph the device from three standard angles: front, back, and charging port. Any modern smartphone camera works. Kitar's upload UI prompts for each angle in sequence so nothing gets missed.
AI scores every damage signal
The vision model reads screen wear, housing scratches, camera lens condition, and port damage from the photos. No serial number entry, no IMEI lookup. The grade is derived purely from visual evidence.
Live market price is applied
Kitar crosses the grade against the current demand signal for that model and condition tier in your city. Prices update daily from verified refurbished listings in Jakarta and Surabaya. You always quote from today's market, not last month's spreadsheet.
Grade card handed to the customer
The result is a printable grade card with the condition breakdown and the offer price. If the customer disputes a grade, the photo evidence is right there. Disputes that used to take five minutes now take thirty seconds.
Grading criteria
What Kitar reads from a photo
Our model was trained on over 280,000 annotated device images collected from buyback desks across Southeast Asia. It reads the signals that actually move resale price.
- Screen wear: micro-scratches, chips, dead pixels, pressure damage
- Housing condition: back glass, frame dents, corner impacts
- Camera lens: scratches, haze, coating damage
- Port and speaker grille: oxidation, debris, physical damage
- Battery health: estimated cycle count from visual cues and reported health
- Accessory completeness: charger and original box when present
API integration
Connect Kitar to your existing systems
The Kitar REST API lets you push intake photos directly from your POS, inventory, or CRM system and receive structured grade and price data back in milliseconds.
POST /v1/grade
Authorization: Bearer <your-api-key>
Content-Type: application/json
{
"model": "iPhone 13",
"storage_gb": 128,
"images": [
"data:image/jpeg;base64,..."
],
"market": "jakarta"
}
Response 200:
{
"grade": "A-",
"confidence": 0.94,
"signals": {
"screen": "minor_wear",
"housing": "clean",
"battery_health": 87
},
"offer_price_idr": 3200000,
"market_updated": "2026-07-21"
}Questions
How the grading works in practice
-
Across our validation set of 12,400 devices graded by both Kitar and experienced buyback staff, Kitar agreed with human graders on the final tier 91% of the time. In the 9% where they differed, Kitar's grade correlated more closely with the final resale price, suggesting the human grade was occasionally influenced by factors the model correctly discounts.
-
Kitar returns a low-confidence flag when image quality falls below threshold (blur, heavy glare, or truncated frame). The interface prompts the staff member to retake that angle. We don't suppress low-confidence results silently: you always see the confidence score alongside the grade.
-
Price data is refreshed daily. The model aggregates verified listings from Jakarta and Surabaya refurbished markets and weights recent transactions more heavily than older ones. For high-velocity models (flagship iPhones, Galaxy S-series), price sensitivity to demand shifts is detected within 24 hours.
-
Yes. Supervisors and managers can override any grade with a note. All overrides are logged with the reason and the staff member's ID, so you build a record of where human judgment diverges from the model. That data feeds back into future training cycles on the Kiosk and Chain plans.
-
Currently Kitar grades smartphones only (Android and iOS). Tablet grading is in closed beta for Chain-tier clients. Laptop grading is on the roadmap for late 2026 but not yet available to new accounts.
-
Photo capture plus upload takes 30 to 60 seconds depending on connection. The AI grade and price return within 8 seconds of the final photo being received. In our own lab testing, the complete intake session from picking up the device to returning the grade card runs under 90 seconds.
Ready to grade every intake in seconds?
Photo-in, grade-and-price-out in under 10 seconds. No arguments over condition, no stale price lists.
No commitment. Expect a reply the same day.