Surprising and not at the same time, thanks for doing the AI overlord's work!
Where would you look to increase the value add of generative AI? I see apps pitching themselves as "AI for X" category, but really it's using AI to replace specific steps in a workflow often at a first pass level. So the value is how much time it saves and how usable is that first pass. Tailored full-site copywriting via a single prompt and conversational edits is a big win, but still might only account for 20% of the full website "category" being solved or assisted by AI. Images and colors even less given easy alternatives exist.
Would be curious if you've seen applications that successfully evolved past the "helpful intern" role where you're not left still doing the majority of the work.
"Instant!" and "completely made by AI!" are good marketing hype but are classic feature vs. benefit -- I feel the urge to tout yourself as an AI company is yet again distracting people from how to make these more useful in the real world.
That's an interesting conclusion! Having toyed around with a bunch of these, I feel they do a decent job of spitting out a cookie-cutter site with fixed elements (like you said) so might perhaps be okay for e.g. a one-page portfolio site for someone who doesn't want to spend much time customizing it. (Then again, one could argue that dedicated portfolio showcase sites are better suited for that purpose.)
And agree: I think once existing players find a way to integrate AI in an intuitive way, it'll be massive. By the way, the crew behind Wordpress also just announced a built-in AI assistant for content creation: https://jetpack.com/ai/
So far, it just helps you write within the Wordpress editor. But I can absolutely see Wordpress incorporating some form of AI-assisted site creation and theme/plugin selection/customization in the future.
Yeah, the Jetpack assistant was exactly what I was thinking about.
I guess my conclusion was that the main value add was saving the 5-30 minutes it takes to browse and choose a portfolio theme. But I don't think that's a huge win for using generative AI?
The thing is, even with no-code sites like Squarespace and WordPress, this assumes that the user wants to spend time and has the bandwidth to look over themes, learn how to use the dashboard to install and customize them, etc. Then populate it all with some content. It can still get pretty overwhelming.
With these hyper-quick AI site builders, the user's input and decision boils down to "This is what I want my site to be about" before they get to see a rough first version. Granted, they still need to customize it afterwards but at least they don't start with a blank canvass.
At least that's how I see the value add: a sort of anti-writer's-block for getting a site off the ground.
But the jury is still out on whether there's enough demand for it to justify the sheer volume of generative AI site builders.
I agree that there's some value in getting past the writer's block stage of site design, but unlike ChatGPT or Midjourney there's huge switching costs to editing your creation after its generated. It would be like if ChatGPT blocked copy/paste and you were trying to use it to write an email.
That's a good analogy and I guess I buy the argument. I haven't played around with the post-editing on these sites as much, but I'm sure they're a far cry from existing site builders on that front.
Will be cuious to see some data on how many people actually find these builders useful beyond the novelty factor and end up launching workable sites through that process.
Surprising and not at the same time, thanks for doing the AI overlord's work!
Where would you look to increase the value add of generative AI? I see apps pitching themselves as "AI for X" category, but really it's using AI to replace specific steps in a workflow often at a first pass level. So the value is how much time it saves and how usable is that first pass. Tailored full-site copywriting via a single prompt and conversational edits is a big win, but still might only account for 20% of the full website "category" being solved or assisted by AI. Images and colors even less given easy alternatives exist.
Would be curious if you've seen applications that successfully evolved past the "helpful intern" role where you're not left still doing the majority of the work.
"Instant!" and "completely made by AI!" are good marketing hype but are classic feature vs. benefit -- I feel the urge to tout yourself as an AI company is yet again distracting people from how to make these more useful in the real world.
That's an interesting conclusion! Having toyed around with a bunch of these, I feel they do a decent job of spitting out a cookie-cutter site with fixed elements (like you said) so might perhaps be okay for e.g. a one-page portfolio site for someone who doesn't want to spend much time customizing it. (Then again, one could argue that dedicated portfolio showcase sites are better suited for that purpose.)
And agree: I think once existing players find a way to integrate AI in an intuitive way, it'll be massive. By the way, the crew behind Wordpress also just announced a built-in AI assistant for content creation: https://jetpack.com/ai/
So far, it just helps you write within the Wordpress editor. But I can absolutely see Wordpress incorporating some form of AI-assisted site creation and theme/plugin selection/customization in the future.
What a time to be alive!
Yeah, the Jetpack assistant was exactly what I was thinking about.
I guess my conclusion was that the main value add was saving the 5-30 minutes it takes to browse and choose a portfolio theme. But I don't think that's a huge win for using generative AI?
The thing is, even with no-code sites like Squarespace and WordPress, this assumes that the user wants to spend time and has the bandwidth to look over themes, learn how to use the dashboard to install and customize them, etc. Then populate it all with some content. It can still get pretty overwhelming.
With these hyper-quick AI site builders, the user's input and decision boils down to "This is what I want my site to be about" before they get to see a rough first version. Granted, they still need to customize it afterwards but at least they don't start with a blank canvass.
At least that's how I see the value add: a sort of anti-writer's-block for getting a site off the ground.
But the jury is still out on whether there's enough demand for it to justify the sheer volume of generative AI site builders.
I agree that there's some value in getting past the writer's block stage of site design, but unlike ChatGPT or Midjourney there's huge switching costs to editing your creation after its generated. It would be like if ChatGPT blocked copy/paste and you were trying to use it to write an email.
That's a good analogy and I guess I buy the argument. I haven't played around with the post-editing on these sites as much, but I'm sure they're a far cry from existing site builders on that front.
Will be cuious to see some data on how many people actually find these builders useful beyond the novelty factor and end up launching workable sites through that process.