What we shipped
- The Audit Engine runs an AI Visibility Audit for a single client, built around the personas that client actually cares about, asking the questions those buyers would ask.
- It produces a client-ready audit deck: what the AI assistants say today, where the client is missing, and the specific changes to make. Every prompt is run three times and averaged, and the deck self-checks its own content and visuals before a human sees it.
Why we built it this way
The first AI Visibility Index looked at a whole sector. It taught us what was worth measuring, and it made the next step obvious: clients needed their own version of it. Not a hundred-page sector study, but a focused, version-one audit of how visible they are to AI for the questions their customers actually ask, with recommendations they can act on straight away.
How it benefits clients
- Concrete, implementable changes, not just a score to admire.
- Fast payback. AI crawlers pick up changes far quicker than traditional search crawlers, so the work shows sooner. For one pest-control client, a handful of small content tweaks took them from zero visibility to appearing in every relevant answer.
What's next
More verticals, and deeper recommendations as we learn what moves the needle in each one.
Technical innovations
The averaging is the important bit. Ask an AI assistant the same question twice and you get two different answers, so a single run is noise. We run every prompt three times and take the average, which strips out most of that probabilistic wobble and gives a reading you can trust. On top of that the deck checks itself: automated tests confirm the content and the charts are right before anyone reviews it.
How it fits the agentic journey
This is our REST research-and-audit discipline turned into an agent, drawing on the same data and design language as the rest of the stack. It is also the manifesto's answer to probabilistic output: where a model wobbles, you do not cross your fingers, you measure it repeatedly and average.