AZET Blog

Give Alt Text and Captions Different Jobs

October 26, 2026

As of 2026-10-23 · The AZET team. We operate azet.io and are writing from that perspective. This article is a creative exercise for fun and reference. Its examples are invented, and the suggested AI workflow is an idea to try with a tool you already use, not a statement that AZET provides this capability.

An invented illustration shows a blue paper boat trapped behind a red mug. The caption says, “The shortcut was not a shortcut.” A draft text description repeats exactly that line. Someone who receives only the description has the joke's conclusion but none of the visual setup.

This exercise separates two writing jobs. One text describes relevant visible information. The other adds an editorial comment. AI can propose wording for both, but the person who can inspect the illustration must check what is actually there.

Define the illustration's purpose

Write one sentence about why the image appears. In our fictional essay, it illustrates a failed attempt to move a boat across a cluttered desk. That context makes the obstruction relevant. The mug's decorative pattern may be incidental.

Now list only visible details needed to understand the example: paper boat, mug, blocked route, and the direction the boat faces. Keep interpretations separate. “The boat feels frustrated” invents an inner state unless the artwork explicitly represents one.

The goal is not an exhaustive inventory of every pixel. It is a description suited to the image's role in this particular passage.

Draft from an inspected description

If your chosen tool cannot inspect images, give it a description you have checked yourself. Do not pretend that a text-only response has verified the picture.

Using this verified description of a fictional illustration, draft concise alt text that explains the visual setup. Separately draft a caption that adds an editorial observation. Keep the two complementary. Do not invent objects, emotions, text inside the image, or accessibility results.

A candidate description might be: “A blue paper boat faces a red mug blocking its path across a desk.” The original caption can then carry the joke. The description provides information that the caption alone does not.

Check the pairing in context

Read the surrounding paragraph, the description, and the caption together. If the paragraph already explains the obstruction, the description may need a different level of detail. If the image contains words necessary to understand the point, account for them accurately.

Watch for duplication that adds length without meaning. Also watch for gaps created by trimming too aggressively. “Boat on desk” is shorter, but it removes the blocked route that makes the illustration useful.

Ask a reviewer what they believe the image contributes after reading the text alternative. Their answer can reveal a missing relationship, such as which object blocks which path.

Keep description and interpretation distinguishable

Suppose the model writes “A reckless boat ignores an obvious obstacle.” That is an interpretation dressed as a visual fact. It changes the tone and assigns intent to an object.

You can use playful interpretation in the caption if it fits the essay. Keep the description grounded in the inspected artwork. When an image is intentionally ambiguous, describe the ambiguity rather than resolving it through invention.

Save the final pair with the image version. Cropping out the mug would invalidate the description and the joke, even if the filename stayed the same. A changed image deserves a fresh check of both texts.

FAQ

Does this exercise establish accessibility compliance? No. It is a writing exercise for one fictional image. A complete accessibility review involves the actual content, implementation, and people using it.

Should every caption describe the picture? That depends on its purpose. In this example, description and commentary work together. Other images may need a different arrangement.

Can I accept an AI description without looking? Do not treat it as verified. Inspect the image yourself or have someone who can inspect it check every concrete detail before publication.

A note from the makers

Our product reference is the maker introduction dated October 3, 2026. That document describes azet.io as a portal centered on the azet AI assistant and records an early-access email waitlist. It also says pricing was unpublished at that time. Those are dated statements from that introduction; this exercise does not establish a later launch, a feature release, or a pricing change.

If this kind of careful, everyday work interests you, visit azet.io to join the early-access waitlist. Read the information shown there before registering. We are sharing a way to think through a small task, with the final judgment left to the person doing it.

aiaccessibilityimagescaptionsediting