AI already answers questions about you.
Some of the answers are wrong.
ChatGPT, Claude, and Perplexity all describe your organization to people who ask.
They do it whether or not you’ve ever checked.
Most of it comes from structured data you don’t control and have never seen.
When that data is thin, they guess.
We pull each engine’s answer separately, check it against your real records, and fix what’s wrong at the source.
The Shift
Search moved. Your story stayed put.
For most of the last two decades the job was to rank on Google and write a clear website.
That still matters. It’s no longer the whole job.
Now people type your name into an AI engine and get one answer instead of ten links.
No list to scan. No second opinion.
That answer is assembled from your schema, your Wikidata entry, your citations, and what other sites say about you.
Your website is one input, and not the loudest one.
The Gap
Everyone else sells visibility.
We start with accuracy.
The AI search agencies all promise the same thing: get found, get cited, get traffic.
None of them ask the first question.
Is what the machines already say about you true?
Traffic doesn’t fix a wrong answer. Correct data does.
We asked ChatGPT and Claude what time a national coffee chain closes on Sunday.
Both said 10 p.m. It closes at 11 p.m.
Perplexity checked live and got it right. The other two guessed.
How It Works
Four things, in this order
Entity foundation
An engine can only describe you if it can tell you apart from everyone else with your name.
We build that layer: Wikidata, schema markup, and the knowledge‑graph records the engines actually read.
Citations and authority
Engines trust what other sites say about you more than what you say about yourself.
We find where you’re described wrong or not at all, and go fix it at the source.
Answer coverage
Your buyers ask engines specific questions before they ever reach your site.
We test those questions, find the ones where you don’t appear, and build the pages and data that put you in the answer.
Search that still works
Google didn’t stop mattering. Technical SEO, site structure, and content still carry weight, including inside AI answers.
We keep that running instead of trading it for the new thing.
Then we check all four again on a set schedule, because the engines change their answers without telling you.
What We’ve Built
We own the tools we run this with
Our own entity research tool.
It pulls what the AI engines say about an organization, compares it against that organization’s real records, and lists what’s missing from its Wikidata and schema.
Waterloo Track.
A first‑party, server‑side tracking framework, built because the analytics most teams rely on stopped reporting who’s actually visiting.
It runs today at Duarte, Hope International, and the H.E. Butt Foundation.
One read per engine, never averaged.
We read ChatGPT, Claude, and Perplexity separately. Each one pulls different sources and gets different things wrong.
An average hides exactly the gap you need to see.
We won’t promise rankings or panels.
We’ll show you what’s true, what’s missing, and how to fix it.
Proof
Who we do this for
A real fix
Earlier this year, we asked ChatGPT who founded a national leadership nonprofit we work with.
It named the wrong person.
We fixed the structured data and entity records the engines read.
Ask again today, and every engine gets the founder right.
Duarte
The communication firm, where we work as the fractional marketing strategy team, advising directly with Nancy Duarte and the executive leadership.
Hope International
Microfinance and development programs across 30 countries.
We clarified attribution across web traffic, donor acquisition, and campaign engagement, and we launched and manage their Google Ads nonprofit grant.
The Peach Truck
A seasonal, farm‑to‑consumer brand turned into a national e‑commerce operation.
A rebuilt email automation and audience system helped drive 300% revenue growth and $500K in new personalized email revenue.
H.E. Butt Foundation
We moved the stack to first‑party, server‑side tracking, and consult on campaign structure, messaging, and measurement.
Lonely Ghost
We designed and built their Klaviyo automation from scratch, mapping the full customer journey for a dedicated fan base. Within six months, email revenue grew 9×.
GoNoodle
A coordinated paid and partnership strategy reached 130,000+ students across 600 schools in Los Angeles, generating over 10 million engagement minutes.
Twenty‑five years of this work.
The AI engines are new. Getting the underlying data right is not.
Free Audit
Start with a free AI answer audit
Ask three engines about your organization and you get three different stories.
We’ll pull all three and show you.
You get:
- 1
What ChatGPT, Claude, and Perplexity each say about you right now, quoted, per engine.
- 2
Every claim in those answers checked against your real records,
marked right, wrong, or missing. - 3
Your entity footprint: your Wikidata entry, your schema, and the citations that engines pull from, with the gaps ranked by what to fix first.
It’s free, and it ends with a short report and a call to walk through it.
If you want us in the work after that, we’ll talk about it then.
If not, you still know what the engines are telling your next customer.
You can’t fix what you’ve never read.
Almost every team we talk to has never pulled a single AI answer about their own organization.
Not one, from any engine.
We’ll pull them for you and tell you what’s wrong.