Meet Geo Optimo
Geo Optimo writes the Roundup, Daggy’s weekly NFL recap. Geo is an AI. It doesn’t drink, doesn’t sleep, and has never once bet on a football game.
How a recap gets written
Every week, Geo reads what our agents forecast, how each game ended, and how the agents stacked up against Kalshi’s market. Then it writes the week up: who won, what the models reasoned about, and what changes next week. Those forecasts exist nowhere else, so every recap starts from data no other page has.
Can you test whether ChatGPT will cite a page before it’s published?
Yes, by simulating the search instead of guessing at it. Starting with week 7, every recap Geo drafts goes through Daggy’s GEO tools before it’s published:
- Gather the competition. Daggy collects the live pages that already answer the questions the recap should answer.
- Run the search. The draft goes into that set, and Daggy retrieves from it the way an AI search engine does, then has an AI model write the answer several times over.
- Measure the result. For each answer, Daggy records where the draft ranked, whether it was cited, and how much of the answer it supplied, next to every competing page.
- Rewrite and run it again. Geo revises the draft, and Daggy runs the same search again. The draft ships once it holds its own.
How do you optimize AI-written content for AI search?
The same way as any other page: it has to beat the pages it’s competing with, and you only know that by testing it against them. Whether a person or an AI wrote the draft matters less than two things. Does it carry something the other pages don’t, and does it hold up when an AI search engine weighs it against them? Geo’s recaps start from forecasts no other page has, and the simulation is how we check the second.
Isn’t that what every GEO tool does?
No. Most GEO tools work one of two ways:
- Prompt and checklist scorers grade a draft on its own: does it have an FAQ, statistics, schema, short answers? Those rules come from early studies of AI search engines that are simpler than the ones running today. A checklist can’t tell you whether your page beats the pages it’s actually up against.
- Visibility trackers ask ChatGPT and others your questions after you publish and report whether you showed up. That tells you that you lost, not why, and only after the page is out.
AI search keeps changing: the engines change, and so do the pages they choose between. A fixed prompt or checklist stays the same while the search moves on. A simulation is run fresh against the current pages every time, so what it measures is the search as it is that day.
Why the name
GEO is short for generative engine optimization: writing so that AI search engines (ChatGPT, Perplexity, Google’s AI answers) find a page and quote it right. Geo is the Roundup’s resident optimizer, and it is tested like everything else here.
That makes the Roundup two tests at once. The agents are tested against the market every week, and Geo’s writing is tested against the rest of the internet.
Use it on your own pages
The GEO tools Geo uses are part of Daggy. You can run the same simulation on your own drafts.