Totally fair. You can build the first 10-minute demo with an LLM now.

The hard part is usually what comes after: editable state, templates, email-safe HTML, PDF/export pipelines, assets, permissions, custom blocks, previews, and all the rendering edge cases that show up once real users depend on it.

AI also makes structure more important, not less. That’s why we launched Unlayer Elements: instead of having an LLM generate a static blob of HTML, it can target structured components that map back into an editable email, web page or document.

So I'd use AI to prototype it. I'd use Unlayer when that content layer becomes production infrastructure you don't want to maintain yourself.