Meet Radar: The Agent That Watches Your AI and Search Rankings

Radar is the Vergrank agent that tracks where you stand. Every week it checks two things for each of your pages and keywords: your position in ordinary search, and whether AI answer engines actually cite you. It records both, and that record is the scoreboard every other agent reads from.

This is the third post in our series on the agents that work your account around the clock. Pip the Writer builds your pages and Nora the Gardener keeps them healthy. But both of them are working blind without data: Pip needs to know which topics are worth a page, and Nora can’t tell a slipping page from a steady one without a position history to compare against. Radar is where that data comes from.

What Radar checks every week

Radar clocks in every Monday morning and goes down your list of live pages and tracked keywords. For each one it runs two separate checks, because in 2026 there are two separate places you can win or lose.

• Search position. Radar runs your keyword as a real search and finds where your page lands in the results. This is the familiar “what rank am I” number, the one every SEO tool has reported for twenty years.
• AI citations. Radar puts the same question to four AI answer engines, OpenAI, Anthropic, Google, and Perplexity, and records whether the answer names your site, and where in the answer you appear. This is the number almost nobody else is measuring.

Both results get written down with a timestamp, so over time you don’t just have today’s rank, you have a trend. That history is what makes the rest of the system possible.

Search rank and AI visibility are two different games

Legacy tools treat your Google position as the whole story. That used to be true and isn’t anymore.

When someone asks ChatGPT “what’s the best tool for X,” they get one answer with a handful of cited sources rather than ten blue links to scroll through. If your site isn’t in that short list, you don’t exist for that query, however well you rank in classic search. You can sit at position 3 on Google and still be invisible to the share of your market that now asks an AI first.

So Radar tracks both, separately, and never collapses them into one feel-good number. A page can be winning search and losing AI, or the reverse, and you need to see which. The search check tells you the old game; the AI check tells you the one that’s quietly taking over.

How Radar turns “are we cited” into a number

For the AI side, “did they mention us” isn’t specific enough, because being the first source an answer cites is worth a lot more than being the fifth. So Radar boils each keyword down to a single AI visibility score across those four engines.

The math is deliberately simple. Each engine that cites you near the top of its answer counts for more than one that mentions you in passing, and an engine that doesn’t cite you at all counts for nothing. Add those up across all four and you get a score that rewards being named early and often. A keyword no AI engine has ever cited you for sits at the bottom; a keyword all four put at the top of their answer sits at the top.

That single number is what lets the rest of your team prioritize. A keyword where you’re invisible to every AI engine is a louder signal to write a page than one where you’re already the cited answer everywhere.

How Radar actually runs

A few details about how Radar works, because the value is that these are real measurements rather than estimates:

• It checks the live engines. Radar isn’t reading a cached rank database or modeling a likely position. It runs the actual searches and puts the actual questions to the actual AI engines, the same way your customer would.
• It’s weekly, on your freshest 50 pages. Checking every page against five engines every day would cost a fortune and tell you nothing new, since rankings don’t move that fast. So Radar runs weekly and focuses each run on your 50 most recently published pages, the ones most likely to still be settling. Older pages are stable, and you can always check one on demand.
• It skips what it just checked. If a page was already checked in the last seven days, a regular weekly run leaves it alone rather than spending money to re-confirm the same number. On-demand checks ignore that and run immediately.
• It keeps the receipts. For every AI check, Radar stores the full raw answer it got and the exact sources that answer cited. So when your visibility score moves, you can read the actual answer behind it, not just a number.

What Radar feeds

Radar’s whole job is to be the data layer the other agents stand on. Nothing it records is for its own sake.

• Nora the Gardener reads Radar’s rank and citation history to decide which pages are slipping and need a refresh. Her four refresh rules are all comparisons against the trend Radar built.
• The Keywords team uses the AI visibility score to rank which topics deserve a page next, so Pip writes for the gaps where you’re most invisible, not at random.
• Competitor tracking mines the sources each AI answer cited. When a rival keeps showing up as the cited answer for your keywords, that surfaces as a competitor worth watching.

Without Radar, the rest of the squad is guessing. With it, every other decision traces back to a measured number you can see.

Where to see Radar at work

Radar lives on the Agents page in your Vergrank portal, on the Tracking team. Its checks show up across your dashboard as the rank and AI visibility numbers on every keyword and page, and you can send any single page to Radar on demand if you want a fresh read right now instead of waiting for Monday.

Next in the series we’ll meet Glenda, Radar’s teammate on the Tracking desk, who counts the clicks, impressions, and click-through rates from search itself, the real traffic that lands once you’ve earned the rank Radar measures.

Frequently asked questions

The FAQ above covers the essentials: what Radar tracks, which AI engines it checks, how often it runs, and how it differs from a classic rank tracker. The part worth re-reading is the difference between search position and AI visibility. They’re two separate games now, and a tool that only reports the first one is telling you half the story.