Find out what AI says. Change the evidence. Ask again.

Diagnose

We ask AI what your buyers ask.

What should I use? Who can I trust? How does X compare with Y? Most questions are unbranded, because the buyers you're losing haven't heard of you yet. The comparison questions name you alongside your competitors, because that's where buyers make the final call.

The question set is fixed for the whole engagement. We run it on ChatGPT, Claude, Gemini, Perplexity and Google AI, three times per engine, so a lucky answer never counts as a result.

We track four things separately: whether you're mentioned, whether you're recommended, how high you appear, and whether your pages are cited. No blended score to hide behind. We also record which sources each answer is built from, because that's where the fix lives.

The diagnosis tells you which problem you have:

• Reach: AI can't find you, or rarely does.
• Context: AI finds you, but gets you wrong or files you in the wrong category.
• Conviction: AI knows you, and still recommends someone else.

You see exactly where you stand before anyone proposes a fix.

Execute

If it's reach, we make sure AI can find and use your pages. Pages it can't access get their crawling and rendering repaired. Your most important pages are rewritten so the first lines answer what buyers ask: what you do, what it costs, and the proof that it works. Where no page answers a buyer's question, we build one: comparisons, alternatives, and use-case pages, plus authority articles on Scale. AI can't recommend what it can't find.

If it's context, we give AI one version of the truth. We build a single source of truth for your brand (what you sell, who it's for, your pricing, and where you sit in the category) and make every place AI looks agree with it: your site, structured data AI reads directly, the directories and review profiles already in your answers, and on Scale, Wikidata and the knowledge graphs engines rely on. When every source agrees, AI stops guessing.

If it's conviction, we go where the answer already comes from. We pitch you to the "best tools" roundups AI cites in your category, grow genuine reviews from your customers, and earn mentions on the third-party sites AI trusts. Where Reddit and Quora threads shape the answer, your own experts contribute specific, useful replies, and on Scale, video scripts answer buyer questions in a format AI increasingly cites. AI recommends what others vouch for.

Sprint fixes the foundations: your pages, your facts, and your first placements. Scale builds on them every month, until AI has no reason to pick anyone else.

Re-measure

Then we ask again.

The same questions, on the same engines, run three times each, measured exactly as they were at the start. So when a number moves, it moved because the answer changed, not because we changed the test.

Every report shows which of the four numbers moved, on which engine, and what we changed that month.

On Sprint: we measure every month, and at day 90 we re-run the full diagnosis against your baseline. That's the before-and-after that tells you whether the work paid off.

On Scale: we measure every month and track three competitors alongside you. A shortlist of your most important questions is checked every week, so a drop gets caught early instead of at month-end. Every quarter, we re-run the full diagnosis to set the next quarter's priorities.

AI answers shift day to day. Our measurement doesn't.