CardSight AI Frequently Asked Questions

Question 1

How do I get an API Key?

Accepted Answer

Getting started with CardSight AI is free! Simply sign up for a CardSight AI account at https://app.cardsight.ai to receive your API key. Once you've created your account, you can find your API key in your dashboard settings. Use this key to authenticate all your API requests.

Question 2

What are 'Free' APIs?

Accepted Answer

Free APIs are endpoints that don't count towards your monthly API call limit or purchased API credits. These endpoints are available to all CardSight AI users at no additional cost, allowing you to use certain features without worrying about usage limits. Look for the 'Free' badge on endpoint documentation to identify which APIs are included. While Free APIs do not consume your paid API credits, they still require authentication via your API key.

Question 3

Which SDKs are available for CardSight AI?

Accepted Answer

CardSight AI offers official SDKs for multiple platforms: Node.js/TypeScript (cardsightai on npm), Python (cardsightai on PyPI), Swift (for iOS/macOS development), Java/Android (for JVM and Android apps), and .NET (for C# and .NET applications). All SDKs are open source under the MIT license.

Question 4

Why are some sets split by series while others are combined into one set?

Accepted Answer

When a release has multiple series (e.g., Series 1 and Series 2), CardSight AI looks at the card numbering scheme to decide how to organize the sets. If a set uses the same, continued numbering across both series (for example, Series 1 has cards #1-200 and Series 2 continues with #201-400), it is treated as one combined set. However, if the publisher restarts or changes the numbering between series for a set of the same name, they are treated as separate sets with the series name included (e.g., "Base Set - Series 1" and "Base Set - Series 2").

Question 5

How should I use the ID fields for cards, sets, and releases?

Accepted Answer

IDs are unique handles for looking up entities in the catalog and chaining calls together — for example, find a set, then fetch its cards, or fetch a card's image by its ID. Caching them short-term to satisfy your users' requests is expected and fine. To hold onto specific cards for the long term, the best approach is to save them to a list or collection — those are your own resources, they're stable, and they're built to keep cards. If you're building something that needs a bulk or offline copy of our catalog data, reach out to us at https://cardsight.ai/contact — we're happy to talk through your use case and point you to the right approach.

Question 6

Why are the image and other endpoints free?

Accepted Answer

Endpoints like the image endpoint — and several others — are free because they assume you already made the catalog call that returned the data, so you're not double-charged for retrieving the same information twice. The takeaway: use the catalog to get the data you need, use the free endpoints alongside it, and when you need to persist cards, save them to a list or collection, or reach out to us for bulk or offline needs. See our Terms of Service at https://cardsight.ai/terms for details.

Question 7

Do I need a perfect image to get accurate card identification results?

Accepted Answer

No. CardSight AI is built to handle real-world conditions, including imperfect lighting, slight angles, and cards wherever you find them — sleeved, on a table, at a card show, or anywhere else. That said, a few simple habits like ensuring the full card is visible and avoiding heavy glare will maximize your accuracy.

Question 8

What makes a good trading card identification image?

Accepted Answer

The ideal image shows the entire card, is reasonably in focus, and is taken in decent lighting. You don't need a scanner or a studio setup. A phone camera in a well-lit room is more than enough. The card should fill most of the frame without being cropped, so the AI can see card borders, logos, and design elements.

Question 9

What is an example of a good card identification image?

Accepted Answer

A good image shows the entire card including all edges, with text, photo, and design elements clearly legible. The card should fill most of the frame without being cropped, and lighting should be even without heavy glare or shadows. A bad image is one that is zoomed in too close with no card edges visible — CardSight AI cannot tell if it is looking at a baseball card or just a photograph of a person without seeing the full card borders, logos, and set design elements.

Question 10

Does card orientation matter for AI identification?

Accepted Answer

The card should be upright or close to it. A slightly tilted card is fine, but a card imaged sideways or upside down may reduce confidence. When in doubt, orient the card the way you would read it.

Question 11

Can CardSight AI identify cards in sleeves, top loaders, or graded slabs?

Accepted Answer

Yes. Sleeves and top loaders generally don't impact results. Graded slabs work well too since the API specifically supports graded card identification for PSA, BGS, SGC, CGC, and TAG. The system can detect graded slabs and extract the grading company, grade value, condition, and qualifier.

Question 12

Can I submit an image with multiple trading cards?

Accepted Answer

Yes. The CardSight AI API supports multi-card detection in a single image and will return results for each card identified. Just make sure the cards are not stacked or heavily overlapping so the AI can distinguish each card.

Question 13

What should I avoid when photographing cards for AI identification?

Accepted Answer

Heavy glare on foil or chrome cards is the most common culprit for lower confidence scores. If you are getting a low-confidence result, try diffusing your light source or shooting at a slight angle to reduce reflection. Also avoid submitting very small or zoomed-out images where the card details are not clearly visible — the card should fill most of the frame.

Question 14

What does the CardSight AI confidence score mean?

Accepted Answer

Each identification result includes a confidence level based on how certain the AI is about the match: High (90-100%) means the AI is very confident in the identification, Medium (75-89%) means the AI believes this is likely the correct card, and Low (50-74%) means the AI has some uncertainty about the match. If you are seeing Low confidence results, it is worth reviewing image conditions — better lighting, less glare, or showing the full card can often improve accuracy.

Question 15

Where does CardSight AI pricing data come from?

Accepted Answer

CardSight AI pulls all pricing data from open, public marketplaces. We believe in transparency — you should be able to see the same data we see. Every record includes the source, the price, and a direct link back to the original listing so you can verify it yourself.

Question 16

Why doesn't CardSight AI tell me what my card is worth?

Accepted Answer

Because a single number can't answer that question. What a card is "worth" depends on context — are you insuring a collection, setting a sale price, or deciding whether a deal is fair? Instead of giving you a calculated "comp" or "value" that bakes in assumptions you can't see, we give you the raw market data — auction sales, Buy It Now asking prices, and active listings — and let you decide what it means.

Question 17

What is the difference between auction and Buy It Now pricing data?

Accepted Answer

Think of it like bid and ask. Auction results reflect what buyers actually paid — that is the market clearing price. Buy It Now records reflect what sellers are asking. Together, they give you the spread. You can filter by listing type to see only auctions, only fixed-price listings, or both.

Question 18

Which sports are supported for CardSight AI pricing data?

Accepted Answer

CardSight AI is launching pricing data in beta with Baseball first. Once the data quality is confirmed, additional sports will be rolled out. Card identification and catalog endpoints already support baseball, football, basketball, and hockey — pricing will follow the same path.

Question 19

Why are some cards missing pricing data on CardSight AI?

Accepted Answer

Behind the scenes, CardSight AI's AI-powered matching algorithm connects marketplace listings to the correct cards in the catalog. During the beta period, you may notice listings that the system hasn't matched yet. As the model continues to learn, coverage will improve and more listings will surface automatically over time.

Question 20

What is the difference between the Pricing and Marketplace endpoints?

Accepted Answer

Pricing gives you historical data — listings from closed auctions and past Buy It Now posts. Auction records are completed sales (what buyers paid); Buy It Now records are asking prices. Use it to understand what a card has been listed and sold for over a given time period. Marketplace gives you what is for sale right now — active auctions, live Buy It Now listings, and a direct search link for each card. Use it to see current availability and asking prices.

Frequently Asked Questions

General Questions

Identification Questions

Pricing & Market Data

Still have questions?