The Publish Button Your AI Was Missing
Six months ago we shipped HTML Pub with a one-line description on the homepage and a working product underneath it. The line was "the publish button your AI was missing." The product was a way to take whatever you described to whatever AI you happened to be using and turn it into a live URL with forms and a custom domain in about 60 seconds.
Both held up, the line still does the heavy lifting. People repeat it back to me when they explain HTML Pub to someone else. The product still ships pages in the time it takes to write a paragraph.
But six months in, the people I talk to about HTML Pub now come from totally different starting points than they did at launch. Some show up from ChatGPT, some from a Claude conversation, some from someone else's Squarespace receipt. The line means something different depending on the door they came through.
Here's the doctrine: three things HTMLPub is, three things it isn't, why the distinction matters, and what the next two years look like if we're right.
What HTML Pub Is
A description first publishing layer. You describe what you want, the page is live. The output is a working URL with forms and a custom domain, not a draft sitting inside another tool waiting to be exported.
That distinction matters because everything else in this category is just AI bolted onto a builder. You still open a canvas. You still drag things around. The AI shaves off 20% of the work and you still lose the afternoon. We got rid of the canvas. The product is the prompt and the result
A loop, not a one-shot. You ship a page. You see how it performs, re-prompts, and publishs. The result, three versions in five minutes. Most "AI builders" can produce a first draft but can't be told to fix something specific without falling apart on the rerun. The conversation is the relationship. Your page is something you keep talking to.
The people who get it fastest are already publishing on a cadence. Coaches running the same $497 webinar three times a month. Course creators on a Tuesday drop schedule. Operators juggling a thank-you page today, a sales page tomorrow, a sequel next week. The loop is the point. One-shot generators produce drafts. A loop produces a live operating asset.
AI-agnostic by architecture. We receive AI output through standard protocol. Anything that speaks MCP, anything that produces clean HTML, anything that wants to publish through us can. Claude. ChatGPT. Cursor. Whatever ships next month. The product doesn't have a preferred model and we don't make customers choose one. The publishing layer should be the layer everyone routes through, not another walled garden.
This is the part I'll fight about most. Builders that lock in one model are not building a platform; they're building a wrapper around a vendor. If a competitor's model gets better, your product gets worse. We didn't want that risk and the audience we serve doesn't want it either. Builders should compete on builder quality. Models should compete on model quality. Publishing should be the part nobody owns.
What HTML Pub is not
Not a page builder. I'll say it again because every conversation circles back to it. We don't compete with the editor. We don't compete with the canvas. If you want to drag a column over an image and resize it, you have plenty of options. The fastest one is probably Carrd, the cheapest one is probably WordPress, and the prettiest is probably Squarespace. However, Carrd is an editor when HTML Pub is a description. They design, we actually build.
Not a free version of Leadpages. People keep saying this. Stop. HTML Pub is its own product with its own pricing, its own audience, its own roadmap. Leadpages adds CRO on top: A/B testing, Smart Traffic, lead enrichment, IP resolution. Different job. Different price point. Different person. If you graduate, the upgrade path is there. But you don't graduate by paying more for the same thing. You graduate when your traffic is real enough that optimization moves the number meaningfully.
Not "AI" as a feature. This is the one I'll fight about hardest. AI is not a feature you add to a builder. AI changed what a builder needs to be. The bottleneck moved. Building a page used to take a weekend. Now it takes a sentence. The job changed and most platforms haven't admitted it yet. We did. The whole product is on the other side of that admission.
Why a tagline becomes a doctrine
I picked "the publish button your AI was missing" because every word in it does work.
Publish button. A button. You hit it. The page is live. Not a workflow, not a process, not a "stage." The mental model is one tap.
Your AI. Yours. Whatever you're already using. We don't pick one for you. We don't ask you to learn ours.
Was missing. Past tense. Something was wrong. The category had a hole. We didn't invent a need; we noticed a gap.
When customers explain HTML Pub to other customers in our community or in their LinkedIn comments, they almost always reach for this line or some variant. The line works because the product holds it up. If the product stopped doing what the line says, the line would die. So far, every roadmap argument has come back to this. We add things that make the publish button stronger. We don't add things that turn the publish button back into a canvas.
That's the test. When I'm in a product meeting and someone proposes a feature, the question is whether it makes the publish promise stronger or whether it reintroduces the canvas. Things that reintroduce the canvas get cut, even when they're good ideas, even when customers ask for them. Because once you let the canvas back in, the line is no longer true and the position is no longer real.
The shape of the next two years
The category is going to consolidate around publish-layer thinking faster than people expect. v0 already started shipping a publish flow inside their build product. Bolt is going to ship something similar. Lovable will. Cursor will. The category leaders are going to figure out that the canvas is where their old users live but the publish layer is where their new users come in. They're going to build half of HTML Pub on the side of their main product because their main product can't carry a one-tap publish promise.
That's totally fine. We don't have to be the only product in this lane. We just have to be the one that's been in the lane the longest, the one with the cleanest implementation, the one that doesn't have a 60-person canvas team dragging the product back toward the old shape, and the one whose roadmap is unambiguously committed to the publish promise.
Our job is to keep the line true. Keep the publish button doing what the title says. Keep the loop short. Keep the protocol open, keep the audience clear.
If we keep doing that, the doctrine stays the doctrine. If we ever stop, somebody else takes the line. That's how categories work.
About the author
Michael Sacca is the CEO of Leadpages and the founder of HTML Pub. He has spent the last 13 years building tools for marketers, the last 18 months building tools that route AI output to the open web, and the last six months writing too many homepage drafts trying to get the line right.