Appearance
AI Item Identification
What this guide covers
Using the AI identification feature to get name, description, category, and pricing suggestions for an item — from a photo, a text description, or both.
Prerequisites
- Logged in
- Your ItemChaos server is connected to a vision model server (see Requirements below)
Requirements
The AI feature sends the photo to a vision model server that your administrator runs alongside ItemChaos (a llama.cpp llama-server with a vision model loaded; the server speaks the OpenAI-compatible API). One model reads the photo and writes the suggestion — there is no separate text model.
A suggestion can take a minute or two: the model runs on your own hardware, not in the cloud, and only one photo is processed at a time.
If the AI Identify page returns errors or empty suggestions, the model server is down, busy or unreachable. Contact your server administrator.
Opening the AI Identify page
Click AI Identify in the top nav. The page heading is AI Item Identification.
Getting suggestions
Provide at least one of the following inputs:
Upload Image: Click Upload Image and select a photo from your device. JPEG, PNG, and HEIC formats are accepted. HEIC files are automatically previewed as JPEG via the image preview endpoint.
Description: Type a description of the item in the Description field.
Both inputs together give better results than either alone.
Click Get Suggestions. The page sends the photo and/or description to the AI backend and displays:
| Suggestion field | What the AI returns |
|---|---|
| Title | Proposed item name |
| Description | Expanded description |
| Category | Category string (e.g. "Electronics", "Books") |
| Suggested Price | Estimated market value |
| Condition | e.g. "Good", "Fair", "Excellent" |
| Brand | Manufacturer or brand name if identifiable |
| Tags | Comma-separated tags |
Applying suggestions to an item
- Click Apply to Item.
- A selector appears — choose the item you want to populate.
- The system fills in only the empty fields on that item. Fields that already have a value are not overwritten.
- The AI suggestion metadata is appended to the item's Notes field.
Clearing the form
Click Clear to reset the description field, remove the uploaded image, and clear the displayed suggestions.
Limitations
- The AI is not deterministic — two identical inputs may produce slightly different suggestions.
- Price estimates are rough; treat them as a starting point, not a valuation.
- Brand and condition identification depends on image quality.
- Long or complex descriptions may produce less precise suggestions than a clear photo.
Troubleshooting
"Get Suggestions" returns nothing or an error The vision model server is likely down or busy. Wait a few minutes and try again; if it keeps failing, ask your administrator to check that the model server is running and reachable from ItemChaos.
Suggestions look generic (e.g. "Electronic device") The image may be too dark, blurry, or at an angle the model cannot interpret. Try a better-lit, straight-on photo.
"Apply to Item" fields did not fill in Only empty fields are populated. If the item already has a name, description, and category, nothing will visibly change except the Notes field.
Related
- Editing Item Details — manually filling item fields
- Uploading Photos — adding photos to items