Philatelic.ai Free beta

About Philatelic.ai

A research-built second opinion for stamp grading decisions: what it does, who it helps, how it works, where it's reliable, and what's next.

Philatelic.ai was founded by Ryan Cody, a physics PhD student at Duke University working in experimental AMO physics and quantum computing, with overlap in machine learning. Contact: [email protected] · LinkedIn.

Philatelic.ai is free, AI-powered stamp grading software you use right in your browser: an online app, with nothing to download and no account to create. The model was trained on nearly 250,000 professionally graded examples. It is not a replacement for expert review or certification. It is meant to help you decide which stamps deserve a closer look.

Why use it

Before submission

Estimate whether professional grading is likely to be worth the fee.

Before purchase

Compare the seller's condition description with an independent read.

Before pricing

Use a probability range instead of a single optimistic grade assumption.

Who it impacts

Collectors

Decide what deserves certification, regrade attention, or a closer expert look.

Dealers

Screen inventory and purchases, and give buyers an objective read on condition.

Auction houses

Help route raw material, condition-sensitive lots, and buyer questions.

What it is not

It is not an appraisal, not a certification, and not a guarantee, and it does not replace expert review. It is a decision aid for whether a stamp is worth submitting for grading, how a seller's description compares to an objective read, and what range to price against. The certified grade is the real answer.

Example PreGrade

Suggested: improves accuracy
Most probable grade
98 Superb

Medium confidence

How likely each grade is
Low
70
75
80
85
7%90
41%95
46%98
6%100

We flag some visible faults automatically, but can't detect hidden ones like thins, regumming, or gum disturbance. Absent a noted or detected fault, this estimate assumes the stamp is sound.

Your inputs
Format
Single stamp
Condition
Mint, previously hinged
Country
USA
Scott catalog #
516
Declared faults
None reported

How it works

  1. 1

    Upload a clear scan or photo.

  2. 2

    Add optional details: condition, visible faults, country, and catalog number.

  3. 3

    The model compares the image and details against patterns learned from nearly 250,000 professionally graded examples.

  4. 4

    It returns the most probable grade, likely range of outcomes, and confidence level.

  5. 5

    If there is meaningful downside risk, the estimate is still shown but marked as uncertain.

New to grading? Our guides explain how stamp grading works, why centering drives the grade, and whether your stamp is worth grading.

How to get the best results

Scan, don't photograph

One stamp on a black background. Minimum 300 DPI, 600 DPI suggested, 1200 DPI best. Photos can work if they are flat, sharp, and evenly lit. Avoid glare, shadows, angle, and scanner "enhanced" DPI settings.

Fill in what you know

Condition, known faults, country, and Scott catalog number are all optional, but each one sharpens the estimate. The more accurate detail you give the model, the better the read.

Accuracy

7,461certified stamps used to validate the model
~9 in 10most probable estimates land within one grade of the certified grade, on unseen test scans
<1 gradeaverage miss for the most probable estimate (one grade = one step, e.g. 95 to 98)

Figures are from internal evaluation on professionally graded material, and reflect stamp types similar to our training data; unusual or novel issues may grade less accurately. Everyday consumer scans can also be harder than reference scans; results are reported honestly as more data arrives.

How we think about accuracy

The goal is to be useful without pretending to know more than the input can support. Every estimate shows the full range of likely grades. When the model sees meaningful downside risk, it flags that plainly rather than make the result look more certain than it is.

Limits

Condition issues

Hidden problems, such as thins, regumming, gum disturbance, and some repairs, often don't show in any image, so no image-based grader can see them reliably. Visible faults are flagged, but not perfectly. If you know about a fault, enter it. Sharper fault detection is active model work, not something a better photo fixes.

Used stamps

Cancels and wear make quality harder to judge from an image, so used stamps are a harder case than sound mint material. Broader, more reliable handling of used material is on the roadmap.

Image quality

This is the one you control. The model needs one clear, flat, evenly lit stamp. Glare, angle, shadows, and clutter all hurt the read. Following the scanning guidance above is the single easiest way to improve your estimate.

Roadmap

  • Bulk upload for dealers, auction houses, and larger collections.
  • User accounts so you can save scans, inputs, and past PreGrade results on the site.
  • Model improvements for broader coverage, harder image conditions, and better handling of used material.
  • More data-driven tools for philately.

Frequently asked questions

Can a tool predict a stamp's grade?

It can estimate one. Philatelic.ai reads a scan and returns the most probable grade, the full range of likely grades, and how confident it is, based on patterns learned from roughly 250,000 professionally graded stamps. It is a decision aid, not a certification.

How accurate is the grade prediction?

In testing on 7,461 professionally graded stamps the model had never seen, its most-likely grade landed within one grade of the certified grade about 88% of the time from a scan alone, and up to about 93% when fuller data is available. The average miss is under one grade. Accuracy is highest on sound, fresh stamps; hidden faults like thins or regumming don't show in a photo, so no image-based tool can catch them, which is why every estimate shows the full range of likely grades, not a single number.

Is the stamp grading tool free?

Yes. The PreGrader is free to use during the beta, with no account or sign-up required.

Can it detect thins, regumming, or gum disturbance?

No. Hidden problems such as thins, regumming, and gum disturbance often do not show in any image, so no image-based grader can see them reliably. Visible faults are flagged but not perfectly; if you know about a fault, you can enter it to sharpen the estimate.

What image quality does it need?

The model reads best from one stamp on a black background. Minimum 300 DPI, 600 DPI suggested, 1200 DPI best. Photos can work if they are flat, sharp, and evenly lit, with no glare or shadow.

Is this an official appraisal or certification?

No. It is not an appraisal, not a certification, and not a guarantee, and it does not replace expert review. It helps you decide whether a stamp is worth submitting for grading and what range to price against. The certified grade is the real answer.

Have feedback on a result, or a stamp the PreGrader should handle better?