Why we stopped saying "AI"

Greetingify makes greeting images. You pick an occasion — Diwali, a birthday, a plain Tuesday morning — add a name, and get back an image with a line of text on it, ready to send on WhatsApp. That is the whole product.

For most of its first year, that is not how we described it. The homepage title read "Greetingify — AI Festival Wishes, Good Morning Messages & Greeting Images". The line underneath called it a free AI greeting generator that made AI images. The signup page said it. The pricing page said it. So did the occasions page and the screen where you actually create the greeting.

On 29 July I took all of it out. Eight files on the website, one commit, an afternoon. The Android app followed the next day.

Nothing about the product changed

Same models. Same pipeline. The same images coming out of the same machinery. If you had used Greetingify the week before and the week after, you would not have noticed a single difference in what it produced.

What changed was the claim we were making about why anyone should care.

The word was answering the wrong question

Someone opens Greetingify because Diwali is on Thursday and they have fifty people to wish, and they would rather not forward the same picture that is already going around three WhatsApp groups. That is the need.

"AI" says nothing about it. It answers a different question — how was this made — that the person never asked. And I had written it there because, at the time, that was the part of the work I found interesting. The models were the hard bit. The hard bit felt like the important bit. It is a natural mistake and it is almost always wrong.

Worse, the word carries a suggestion I did not want anywhere near this product. "AI-generated" tells people the output is machine-made, and therefore, somewhere in the back of the mind, generic. That is the exact opposite of what someone should feel about a greeting they are about to put their own name on and send to their mother.

There is also the plain fact that the word had stopped carrying information. When every app on the store says AI, saying it puts you in the crowd rather than apart from it.

What it says now

The title became "Greetingify — Personalised Festival Wishes, Good Morning Messages & Greeting Images". The generator became personalised greetings. AI-generated images with your name became greeting images with your name. Small edits, all of them, and none of them cost anything to make.

The technology still gets named where naming it is genuinely useful: the page that explains how the product works, and articles like this one. A person deciding whether to trust the thing may well want to know. A person with fifty Diwali messages to send does not.

The test we use now

Every line of copy gets one question: does this sentence describe something the reader wants, or something we did?

Powered by AI is the second kind. So is built on a scalable architecture. So is most of what ends up on a homepage, because the person writing it is the person who built it, and they are still standing inside the work. The reader is outside it, and their interest stops at the result.

What it cost

People do search for "AI greeting maker". Taking the word out means giving up some of those visits, and we knew that going in. We took the trade because a clear product description is worth more to us than a search term, but it is a bet rather than a certainty, and it is one we are still watching.

If the traffic tells us otherwise in six months, the honest thing will be to say so here.