Who is Flygrade?

I am excited that you have stuck around for our second newsletter posting. Every other week I want to talk a bit more about Flygrade (Flygradeit.com) and dive in to what we are working on, what we have accomplished, and what did not work! In the first newsletter, I wrote about the Impact of AI on Card Grading. Flygrade is on a mission to make trading card grading immediate and affordable. To do that - we utilize AI and machine learning which powers our mobile application.

Why was Flygrade created?

It starts with the founders - James Reid and myself, Kyle Kemp. Both of us have been collectors since we were kids. Anything from sports, Pokémon, and Garbage Pail Kids trading cards. We were looking for an opportunity to invest in something together and knew that we wanted to formulate a start-up to build in our spare time. Originally, the idea was to build an exchange/marketplace.

If you are familiar with network effects (and if not check this article here by Tim Stobierski) then it is easy to see why we chose to build a tool versus a marketplace. Once we identified our industry, sports memorabilia and within that industry specifically trading cards, we set out to determine the strategy. Building a tool only requires a network effect of 1. A user can find value in using a tool, such as an application that can pre-grade their trading cards. This was much easier to build than an exchange where at any given moment you would need more buyers or more sellers to keep the user engaged.

Flygrade was incorporated in summer of 2022.

Here is a note I received from James when the EIN cleared. "This is how it starts! There is competition with every product that exists. We focus on user experience in app, proper positioning, and patience with timing."

Why the newsletters about AI and how does it fit in with Flygrade?

Flygrade uses a combination of Artificial Intelligence and Machine Learning.

AI is used in the following ways

  • Card segmentation - The AI looks at the card and we created a segmentation network that separates card images from background.

  • Defect detection - The AI analyzes corners, edges, centering and surface to look for defects. It parses the image out via the segmentation and breaks it up in smaller bits to assess each area of the card, front and back. We are considering building a heatmap of cards graded by our AI, called an interpretability technique.

  • We have fed over 30,000 images in to our neural network. Essentially creating a reference point of cards that have been graded. The machine learning component then API calls this dataset to compare what it is seeing when a card is scanned. A grade is then produced.

  • We are consistently testing our accuracy to our grading rubric to what the algorithm is producing and in future weeks I will share a published report. This will be in the form of a confusion matrix

What has Flygrade accomplished to date?

In our short existence we have spent majority of the time #BuildingInPublic and getting ready for the app release. The app went live on 12/15/2022. Here are a few highlights

  • 3500 hours in app development

  • Nearing 1000 followers across social (please follow!)

  • Selected to be featured on Startup Founder Daily

  • Graded over 200 cards

  • Received 200 downloads in the app store

In the coming weeks I will continue to highlight our company and introduce the Flygrade team. Questions? You can visit our website, contact us right through the app, or email me at [email protected]

Click below to visit our website and download the app!