When things calm down a bit on how components are built, what they can do, and complete documentation, I’m thinking of building out a custom GPT (which I’ll share) for creating Elements components from text descriptions. I’ve not fully thought through the details but generally speaking:
Open GPT to describe what you want a component to do.
The GPT should ask for more information if it needs it to fully understand your ask and meet whatever coding needs Elements components have.
Generate complete code (HTML, CSS, Javascript, perhaps Python) for the desired component.
Explain everything it does in the code.
You copy/paste the resulting code into the custom component interface and test it.
Using screen shots and further text (voice?) interactions, adjust code as needed to get finished result. This includes any error messages.
Yes, it would be cool if Elments had this built in, but I think that is out of scope for the current release track…understandably. But maybe this will help shrink the gap between Elements and the growing number of GenAI website development tools.
If anyone has any suggestions or wants to assist, let me know.
Now, if you want to pre-prompt the whole thing and build it right into to Elements…I think most would prefer that!
In lieu of a great wizards of Realmac releasing something officially, I will create the GPT as soon as I can get access to the docs for a stable custom component structure!
I may have to take the time to do some web scraping. The problem is that I’ll have to do it fairly regularly to keep up with any changes.
Any chance you have all of the dev docs in a text file(s) somewhere. That would greatly cut down on the work involved. I’ve had a lot of success with people using the Tap Forms database custom GPT. I was able to download manual files and train the GPT on them. Now it does a very admirable job of helping with everything from basics to advanced javascript scripting of the database.
It would be great if Brendan would open up his database so that I can use it locally for automations like I use AirTable for SaaS automations. *Yes, off -topic.
I can point the GPT to the on-line docs, but that is less reliable. It would save me a large amount of time and have the benefit of having access to the latest user document changes, but again, not as reliable.
Besideds, I’d like to think that the custom GPT is a place-holder for a future built-in capability using Apple Intelligence to drive an intelligent agent function that helps users (a AI powered Help system) or is a true agent and actually does things within Elements to build websites. In that case, to support Apple’s privacy priority it would need to be able to work offline–like Elements does today. That will require local knowledge files!
I’ve work on a few things and see if I can get it to work.
