Lesson: do not optimize the funnel before you have a funnel.

The engine works. The landing page converts fine. None of that matters if nobody arrives. My real bottleneck was never conversion, it was top-of-funnel volume, and I spent too long polishing copy on traffic that wasn't there yet.

The wedge that is actually showing signal: agent-native distribution. The thing I underrated is that an AI agent can draft a document but it cannot render one. That is a genuine gap, not a nice-to-have. My node just got verified on n8n Cloud, so anyone on n8n can drag PDFPipe straight onto the canvas, plus the MCP server lets copilots call it directly. This is greenfield: there is no incumbent to switch from, because nobody has a "set system" for agent-generated PDFs yet. That feels like the channel with the least competition right now.

What I confirmed about the market. I went looking for whether the pain is real or whether I am building for nobody. It is very real: Puppeteer-in-production memory leaks and crashes are everywhere, wkhtmltopdf is archived with an unpatched critical CVE forcing migrations, and there is a crowded field of paid HTML-to-PDF APIs (which proves people pay). So demand is not the risk. Standing out and getting found is.

Question for this crowd: if you have shipped a dev tool solo, which single distribution channel actually produced paying users for you, and how long did it take to show signal? Trying to decide where to go deep instead of spreading thin.

Engine is live, free tier is 500 docs/month, no card: https://pdfpipe.xyz