From Screenshot to Software

Recently, I noticed something strange while working on software changes for a client.

The client would send feedback through Slack messages, screenshots, or a short video.

I would look at it briefly, take the screenshot, and send it directly to AI.

Then AI would read the conversation, understand the context, find the relevant part of the software, and make the change.

Very often, I did not need to rewrite the feedback.

I did not need to create a detailed ticket.

I did not need to translate the client’s words into technical requirements.

The screenshot was enough.

The old workflow looked like this:

Client feedback
→ understand it
→ rewrite the requirement
→ explain the context
→ build the change

Now it increasingly looks like this:

Client feedback
→ screenshot
→ AI
→ working software

Of course, I still need to check whether the interpretation and implementation are correct.

But the translation work in the middle is disappearing.

A screenshot is no longer just an image.

It contains the interface, the wording, the surrounding conversation, and often enough context for AI to understand what the client wants.

This feels like a fundamental change in software work.

Maybe the future of software requirements is not a perfectly written ticket.

Maybe sometimes it is simply:

“Here is the screenshot. Please fix it.”