AI agents read your site as raw HTML, and you can’t see them. Onto serves them clean Markdown at the same URL, while people and Google get your page as before, and shows you which agents came. Plus an AIO score, and a Read API and MCP for builders.
Hi Product Hunt, I'm the maker of Onto.
AI agents now read the web for people. Ask ChatGPT or Claude what a tool costs and it fetches the pricing page. The trouble is what it gets back: Stripe's pricing page is 699 KB of HTML, and the same page as Markdown is 17 KB. The rest is styles, navigation, SVG icons and attributes. An agent that can't find the answer in all that still answers. And you never see it happen, because most agents never run your analytics.
Onto fixes that from the site's side:
→ Serve SDK for Next.js: at build it writes a Markdown copy of every page. GPTBot, ClaudeBot, PerplexityBot and the rest get the copy; people and Googlebot get your page exactly as before, from the same URL.
→ See who's reading: which agent came, which page it read, and how much lighter the answer was.
→ Tell them what matters (Pro): add context to a page that only agents receive.
→ The AIO score: starts at 100 and takes points off for named faults you can fix. The scan is free, no signup.
→ Building agents instead? The same engine is a Read API and an MCP server for ChatGPT, Claude, Cursor and others.
Every number on our site is measured and dated. Our own pricing page is 79.4 KB to a browser and 5.7 KB to GPTBot.
Free to start: Serve on one site, Read with 1,000 credits a month, no card.
Next is Act: agents that do things on your site, not just read it. I'd love to hear what you'd want it to do. And scan your site at buildonto.dev and tell me your score, I'll answer every one.
Report
Hunter
It's the same content in a different format, at the same URL: content negotiation. The Markdown is your page's own text, and Googlebot isn't on the SDK's list, so search sees your HTML exactly as before.
Report
Hunter
They're readers for people building agents. Onto's core is on the other side: your site answers agents itself, and you see which ones came. We have a Read API and MCP server too, for the builder side.
Report
Hunter
Next.js 14 and up on the App Router today. Vue, Angular and WordPress are on the roadmap.
Report
Hunter
No. Googlebot gets your HTML as it always has. Google-Extended and GoogleOther, Google's AI crawlers, get the Markdown.
It's the same content in a different format, at the same URL: content negotiation. The Markdown is your page's own text, and Googlebot isn't on the SDK's list, so search sees your HTML exactly as before.
They're readers for people building agents. Onto's core is on the other side: your site answers agents itself, and you see which ones came. We have a Read API and MCP server too, for the builder side.
Next.js 14 and up on the App Router today. Vue, Angular and WordPress are on the roadmap.
No. Googlebot gets your HTML as it always has. Google-Extended and GoogleOther, Google's AI crawlers, get the Markdown.