How Origo works.
Two stages. Measure, then close. Here is exactly what happens.
One URL.
That's all we need. The engine reads every page, in a real browser, including the text sitting inside images that most crawlers skip entirely.
Out of that comes your knowledge base, your competitor set, your ideal customer profile, and a prompt set built from the questions your buyers actually ask. It also tells you which AI crawlers your site currently blocks, and how much of your content disappears for engines that don't run JavaScript.
Nothing is invented. Everything the engine knows about you came off your own site.
Every day. Every engine.
Your prompts run daily across the engines in your plan, and the engine reads the full answer that comes back.
For each one it records whether you were named, where in the answer you landed, how the engine described you, and which competitors were named instead.
The headline number is citations per prompt. For each question that matters to you, how often and where you come up. It beats a single blended percentage, because it is tied to the exact questions your buyers ask, not an average across questions nobody asks.
Every gap comes with its fix.
A gap you can't act on is worth nothing. So every gap comes back with the solution to close it.
The content brief tells you what to write, how to structure it, how long, and which competitors are winning the answer you want. Press generate and the finished piece is written for you, in your voice, from facts already on your site. Schema block. llms.txt line. Ready to publish.
Every fact comes from what the engine learned about your business. No invented numbers, no invented customers, no claims you'd have to walk back.
You publish. We check.
You publish the work yourself, on every tier. Mark it published, and every run after that watches for it.
When an engine starts citing the page you shipped, you see it on the next run. The loop closes, or it doesn't, and you find out either way.
One engine decides whether a gap is confirmed.
Perplexity, Google AI Overviews and AI Mode search the live web every time. Gemini grounds its answers and shows its sources. ChatGPT is the variable. Most of the time it searches first. Sometimes it answers from memory.
That single difference is what the labels read.
Actionable. The engine searched the web, came back with sources, and still didn't name you. Confirmed. Act on it now.
Monitor. The engine answered from memory this time. The other engines confirm the gap, so it's almost certainly real, but we won't call it confirmed until it is.
A recommendation usually covers several questions at once. If any one of them comes back confirmed, the whole recommendation is Actionable, and the card tells you how many. Confirmed live on 3 of 5. Only a recommendation with nothing confirmed sits in Monitor, and the next day's run gives it another chance.
Daily isn't about a fresher number. It's about how fast an unconfirmed gap becomes one you can act on.