The Nine Tools We Built to Get Brands Cited by AI
Every instrument we built for AI search visibility, what each one does, and where it sits in the method. All of them are internal now — here is what they are.
7 min read
We built nine tools. For a while some of them were public and anyone could run them. They are internal now, and this page is what replaced them.
The short version of why: every public run cost us real money in tokens, and the people running them were not the people we wanted to talk to. Running them ourselves, in bulk, on brands we actually want to work with, costs a fraction of that and produces something far more useful — a report a founder can act on rather than a score they screenshot and forget.
So instead of letting you run them, we will run them on your brand and show you what came back.
Here is every one of them, in the order we actually use them.
First, find out what the models think you are
LLM Perception → — We ask the engines what they know about a brand from training data alone, with no web access, then compare it against what the brand's own site says. The gap is the finding. Scored out of 100 across Recognition, Accuracy and Positioning. Nothing downstream matters if this comes back low.
AI Readiness Audit → — A structural diagnostic of whether the engines can read you at all: crawl permissions, JavaScript rendering, schema completeness, entity consistency, content structure. Runs in about a minute and decides what the first ninety days contain.
Competitor Scorecard → — Grades a brand against three competitors on AI-search presence, content depth, SEO maturity and brand clarity. We pick the three, and picking the right three is the most consequential judgement in the exercise.
Then, grade what already exists
Pillar Analyzer → — Grades the content a site already has: which subjects it genuinely owns, which it touched and abandoned, which have volume but no structure holding them together. Most brands find one accidental pillar and two abandoned ones.
AIO Article Checker → — Grades a single article for whether an engine can lift a claim from it and attribute it cleanly. Meta, structure, schema and content, weighted, out of 100. Ranking and being cited are different jobs.
Then, decide what to build
Pillar Cluster Engine → — Reads a site and proposes three to five pillar pages with six to ten clusters each, ordered by what to ship first. No search volume, no SERP data, no competitor sitemaps. Every title has to work at the category level.
Review Mining and Sentiment Gaps → — Pulls reviews from the open web, not just a connected reviews app, and looks for the distance between what a brand claims and what customers keep saying back.
Then, produce
The Reddit Citation Engine → — Finds threads the engines are already citing where the existing answers are weak, and puts the client's own voice in them. Two to four placements a month, maximum. We do not operate accounts on your behalf.
The Article Production Agent → — Every article traces back to one of four kinds of evidence: strategy, reviews, Reddit, or the pillar map. Nothing gets written from a blank page.
Some of these are finished software. Some are steps our team runs by hand with tooling that keeps changing as the models do, and we say which is which on each page rather than pretending otherwise.
Your free AI Perception Report is the first three of these, run on your brand, before we ever speak.
