Posted by kevlened 3 days ago
For example, creators behind libraries like Tailwind could sell Claude skills or MCP server solutions.
If I could pay $20 to make my AI agents significantly better at writing state-of-the-art Tailwind code — while knowing that my purchase directly supports the Tailwind community and its long-term sustainability — I would happily do so.
They could sell training data too. Though, UIs are relatively solved. But great UIs and criticizing UIs aren't.
Learned a lot from Refactoring UI, and I know (from trying) that it's impossible to make a code review bot based on out of the box sota models today. Vision capabilities are lacking here, and I can see demand for more data here. And Adam's taste likely fits well here.
This tells me the problem wasn't AI but the overall business wasn't healthy. Docs don't drive sales.
It’s unfortunate that google helped kickstart the world wide web but now they’re extracting everything while polluting search results with ads
Where on earth did you get that idea? The web existed long before Google - Google just found a unique way to monetise other people’s content
Why pay for a template when AI's can shit out your entire design system and multiple templates in 5 minutes, not to mention competition from other template systems like shadcn that are completely free.
And yes they might not be the best quality but you just prompt it until you like it and then use it as a reference.
LLMs, or Tailwind. Pick one!
@adam: this is just an idea. Have you tried reaching out to OpenAI, Anthropic et al to become sponsors of tailwind? Could that be a viable revenue path?
Maybe you could offer LLM friendly docs to them, or access to something valuable for them? Or maybe they’re just happy to sponsor.
Tailwind and its popularity make LLM’s more valuable, so I’m sure the model makers want Tailwind to thrive.
Any other monetization ideas to help Adam?
Oh my days, how cringeworthy.
- The value they created (mindshare, shared “standards” for naming properties, and design atoms) and what they charged for (templates that AI can replace) are two different things — and AI has shortened the time it takes for this discrepancy to show up.
- Isn’t almost all of Tailwind’s value actually in that shared semantics (“mt-2” = a small top margin) — not only in users’ heads, but now also in LLM training data? Isn’t it more of a standards organization (like ISO) than a product company (yes, sure, standards are also a product/service)?
- They criticize AI for extracting value, but I wonder if Tailwind's business model is also value extraction from the standards they established.
- And isn’t it almost a miracle that a token library and the idea of “let’s name five margin sizes” (which they weren’t even the first to do - I started with Basscss) could sustain an ~7-person company for so long?
I tried this LLM prompt for deep research: "Tailwind is laying off people. I consider their business much more of a standards body (like ISO) — their main value is the mindshare and shared semantics and design atoms. What business models could they adopt from standard bodies’ business models?"
However, after reviewing the suggestions, I believe tailwind movement is probably not large/important enough to make money in a similar way (sell certification, membership with governance privileges, training ..).
Two interesting ideas: "Keep human docs free, but put machine-optimized “spec corpora” behind licensing (because AI is the channel disrupting them)."
"Stop relying on docs-as-marketing if AI is eating that funnel, and instead monetize the privileges and assurance around the standard (governance, certification, conformance, canonical distribution)."
(Don't get me wrong, I love using Tailwind, but I believe they need to see their business realistically.)
Real shame, and I fear it is just the start of the impacts of AI on our industry.
"I am happy to share that we (the @GoogleAIStudio team) are now a sponsor of the @tailwindcss project! Honored to support and find ways to do more together to help the ecosystem of builders."