Prompt Dress 0.2.0 - Release Notes

Prompt Engineering is dying Schrödinger’s death. On the one hand, the demand for prompt engineers in the applicable job markets remains unabated. On the other hand, …

will likely eliminate the need for sharing, selling, saving and reiterating prompts as we did until now, since “primed” models will now become “natively” reusable. OpenAI is not the first company to use this idea. OpenAI’s board member Adam D’Angelo is notoriously behind Poe, a startup with much of the same idea.

For me and this project, this means that this way of organising and keeping prompts will likely have little merit in the future. This is why I will be shifting my attention. The Prompt Dress browser extension will enter maintenance mode for the foreseeable future since the data Google provides suggests that a surprisingly high percentage of the users that installed it use it regularly.

First, once your browser updates to the next version of the extension, you will find a new entry in your “context menu” (when you select some text and right-click).

This means that avid users of the extension will probably be collecting prompts even faster than before, which means that searching, filtering and sorting prompts needs to improve, so I included an all-new filter panel in the same release.

You can now search, filter by tags, or hide the prompts from the Dresses view that only exist as part of a Shawl.

I also redesigned the Dress cards so you can now see which Shawls a Dress is part of in the Dresses view.

There was a single issue reported to me this whole time, which was that if a Shawl consisted of more than three Dresses, there was a possibility the chain-pasting broke. I was able to reproduce and fix the issue, including a complete refactoring of the whole mechanic to be (almost) completely asynchronous, which should come with some overall performance improvements.

The topic of productive human interaction with AI is still fascinating to me. While AI capabilities are becoming crazier by the day, the question remains: How can we, the average consumers, leverage this power to our advantage most effectively? While “Prompt Dress” is, in hindsight, somewhat of a naive approach to this topic, I believe that the question warrants ever deeper exploration.

This is why, on the one hand, I have released this curated list of custom GPTs, which I have built in only 24 hours (you can read about that here). On the other hand, I plan to experiment with different ideas publicly, here on my substack newsletter, my blog over on Medium and my YouTube channel.

Of course, this is more fun as more people participate, so I ask you to subscribe to those if they interest you and share them with people in your circle who might also be.

Share Prompt Dress

You will hear from me as usual, not too soon, so I will probably be the first to wish you a Merry Christmas and a fulminant start to your 2024!

Konstantin