Help & Guides

Step-by-step guides for every tool, recipe, and setting

Getting StartedThe first hour — a guided tour

The first hour — a guided tour

The first hour, from a real account

Every screenshot below comes from a genuinely new account walking the same path you will: no keys at the start, no projects, a site the crawler had never seen. Three steps, in this order — the first one needs nothing from you but your web address.

The dashboard greeting a new account with the three steps
The whole method on one screen: check your site, add an AI engine, start closing the gaps — and the one thing that is deliberately NOT a step: connecting Google waits until you are ready to publish.

Step 1 — check your site

Type your address and press "Check my site". The check reads your site from the outside, the way an AI crawler does — no passwords, because a stranger doesn't have your passwords and a stranger is exactly who we're pretending to be. You get one list, worst first: the order is the strategy.

The site check findings list
A real first check: 9 things to work on, ordered by what matters. Note the honest row at the top — whether AI names you is marked NOT MEASURED YET, because nothing has been asked yet.
An expanded finding with its paste-ready fix
Every finding opens into why it matters and the exact fix — most hand you the block to paste. If someone else runs your site, forward them the card.

Still step 1 — correct what we think you do

The same crawl reads your pages and proposes what your business is about. Everything on that screen is a proposal, not a decision: each topic shows the page it was found on, the business type is a question with our answer pre-selected, and nothing is saved until you press Confirm. Sometimes the honest answer is that your site says too little to tell — and the app says so instead of guessing:

The derivation screen reporting too little to tell
A single-page site gave the crawl almost nothing — so it refuses to invent topics and asks you to add them. Type what you want to be known for, in the words a customer would use.

Step 2 — add an AI engine

The check reads your site; it can't read AI's answers without an engine to ask. The app literally asks the engines the questions your customers ask — on your own key, your own quota, your own data. Gemini has a free tier, which is why it's the suggested first engine. Each engine you add is another one you can see into.

AI Settings with providers configured
AI Settings after adding two providers. The fallback order means a failing provider never silently stops the work.

Measure where you stand

With topics confirmed and an engine connected, the app offers a first measurement — and tells you exactly what it will and won't establish before it spends anything:

The baseline offer stating its question count
The cost and the caveat, stated up front: a handful of questions is a first look, not yet a baseline. Confirm more topics and it earns the word.
The finished first measurement
A real first result: mentioned in every answer, recommended in none, cited by 13% — each figure with its confidence interval, and 'fix this first' pointing at citations. A low first number is normal; it is the starting line, not the verdict.

Your topics, and the four switches

Everything the app measures and builds hangs off your topic list, on the project page. Each topic carries four independent switches, each a separate spend: Measured (we ask the AI engines about it — your AI key), Building (the engine creates and publishes evidence for it), Watching (Google's AI Overview is checked — needs a DataForSEO key), and Ranking (a classic nightly position check). Any of them can be off while the others stay on.

The topic list with four switches per topic
One list, four switches per topic, and the cost of each stated in the paragraph beneath — every measured topic is prompts on your own key, every run.

You have your first numbers — now what?

This is the moment most tools leave you hanging, so here is the answer plainly. 1. Read one number first: CITED. Mentioned means the engines recognise your name; cited means they point people at your site. Moving cited is the work, and the score card's "fix this first" already points at it. 2. Start the engine (Step 3). It works on exactly what the measurement found missing — finds the evidence gaps, builds the pages, publishes them, and re-checks on every cycle. Everything it publishes leaves a link you can open. 3. Know where your numbers live. Reports & Tracking → Visibility Audit holds every run; run the check at least twice and the trend view starts, engine by engine. 4. Then climb the ladder, one rung a week: fix the check's findings top-down · connect your own site for publishing (a page you own outranks one on a Google domain) · connect Google when you're ready to publish (it also brings your Search Console data in, free) · add a second engine · add the Serper key for rankings and live discussions · and when you want it hands-off, move to a plan where the engine runs on its own schedule. 5. The weekly habit: read the dashboard's "Only you can do this" band with your coffee, glance at your topic list, and open the Monday report email if you've turned it on. Everything else is the engine's job.

The app will tell you "not enough data yet" and "too soon" — in those words. That's deliberate: every number here is one you can defend, and a tool that invents certainty is selling you noise. When it says wait, waiting is the strategy.