AI for Comparing Typography Specimens
As of 2026-10-16 · the AZET team. We operate azet.io and publish this guide for fun and reference.
Two printed headings can look equally appealing until you add the line that actually matters: a long room name, a tiny caption, or a date with several similar digits. For this fictional exercise, imagine choosing lettering for a small reading club's notice. There are two unnamed specimens, A and B. Nobody has tested them yet.
We would start with the notice, not a request to name the better typeface. If you use an AI text assistant, give it your observations and ask for a comparison plan. The seeing, printing, and final choice belong to you. A text description is not a substitute for looking at the specimens.
Build one page both candidates must survive
The hypothetical notice needs a heading, a room label, a short paragraph, and a time. Use identical copy in both specimens. Otherwise, an attractive word in one sample can distract from the different task you gave the other.
Here is invented specimen copy for this exercise:
- Heading: Lantern Reading Circle
- Room: Willow Hall, Lower Level
- Time: 11:10 on Thursday
- Paragraph: Bring a short passage to share. Listeners are welcome, and reading aloud is optional.
Record the intended role beside each line. The heading is for recognition; the room and time carry practical information. Those roles make the comparison less vague. You might accept a distinctive heading while deciding that a particular time line deserves another look.
Turn impressions into observations
Suppose the fictional reviewer writes: "A feels friendly. B looks serious. A's room line wraps after Hall. I reread the time in B. Both headings fit."
Only part of that note describes a visible event. Friendly and serious are personal impressions. Wrapping and rereading refer to something the reviewer noticed. Keep both kinds, but label them separately.
A usable prompt would be:
Organize these fictional specimen notes into observed behavior, personal impression, and unanswered question. Preserve the labels A and B. Do not identify the fonts, infer their technical properties, or choose a winner. For each unanswered question, suggest one comparison I can perform using the same notice text. Notes: [paste your own notes].
An illustrative comparison sheet could contain these rows:
| Note | Category | Next comparison |
|---|---|---|
| A wraps after Hall | Observed behavior | Put both room lines in the intended space |
| B feels serious | Personal impression | Compare that impression with the notice's intended tone |
| Time in B was reread | Observed behavior | Look again at both time lines in the same setup |
This sheet does not establish that B is difficult to read. It records one hypothetical person's rereading and gives that person a next action.
Correct the attractive overstatement
Imagine a proposed summary says, "A is welcoming and more readable, so it is ideal for community notices." That is an invented draft sentence, not a tool result or a finding.
The human correction should be narrower: "For this notice, I prefer A's heading. Its room line still needs a layout check. I reread B's time once and have not repeated that comparison."
Notice what remains unresolved. A preference about the heading has not become a claim about the whole family of letters. One rereading has not become a verdict. The correction also keeps an inconvenient observation about the preferred candidate.
Ask for a second pass if useful:
Rewrite my decision note using only observations I supplied. Separate the choice for the heading from the choice for practical information. Retain unresolved items. Avoid claims about other readers or other projects.
Make a decision small enough to revisit
For the fictional notice, a provisional decision might be to use A for the heading and postpone the room line until its spacing has been checked. Another reasonable result is to keep both candidates and print a revised notice.
Save the actual specimen page with your note. Include which line prompted each judgment. "Prefer A" is hard to revisit; "prefer A for Lantern Reading Circle at the tested size" gives your future self something concrete to inspect.
FAQ
Should both specimens use the same text size?
Start with the same chosen settings and record them. If you later adjust a candidate to suit the notice, treat that as a second comparison. This separates your initial comparison from the practical question of how each candidate looks after your own layout choices.
What if I cannot explain why one heading feels better?
Keep the preference as a preference. Point to the specific word or line, then try describing its spacing, shape, or relationship to the surrounding text in ordinary language. You do not need to invent a technical explanation to justify a personal choice.
Can I compare specimens without sharing images?
You can ask a text assistant to organize your written observations and prepare follow-up checks. It cannot establish visual differences from notes you never supplied. Keep the specimen IDs visible and return to the actual pages whenever a description sounds more certain than your observation.
A dated product note
Our maker introduction, dated 2026-10-03, describes azet as an AI assistant and an email early-access waitlist. This specimen exercise is general guidance, not a claim about an available AZET feature.
For the waitlist described in that dated introduction, visit https://azet.io and check the website for the latest availability.