Live AU accounting & tax

When your customers ask AI who to hire, does it say your name?

Buyers now ask ChatGPT and Perplexity instead of Googling. Citedex asks the answer engines the exact questions your buyers ask — every week — and shows you who gets named, who gets cited, and how to become the answer.

Real engine answers, about two minutes, no credit card.

AI answer share of voice

Share of all tracked-brand mentions across this vertical's prompt set.

Measured across
Loading engines…
Prompts tracked
Answer engines
Answers archived
Honesty is a feature

You always know exactly what was measured

"AI visibility" tools rarely say where a number came from. Every metric we show is labelled by the engine surface it was captured from, and the raw answer is archived verbatim behind it.

synthetic A model answering from weights

The engine's own API replying with no live retrieval. This is what a model "believes" about your market.

api A search-grounded API

The vendor's retrieval-backed endpoint. Answers cite live sources, so publishing can move them.

web The consumer web interface

What a buyer actually sees in the product. Closest to reality, hardest to capture at scale.

How it works

Three steps, run every week

01 Queries

We ask the questions your buyers ask

A curated library of real buyer prompts for your vertical, run weekly across the answer engines — not keywords you guessed at.

02 Impressions

We record who gets named and cited

Deterministic brand extraction — share of voice, presence, position, and the sources each engine trusts. Every raw answer archived.

03 Coverage

You fix the blindspots

See the prompts where competitors appear and you don't, and the exact domains to get cited by to change the answer.

Zero learning curve

Google Search Console, but for AI answers

If you have ever done SEO, you already know how to read this product. The nouns map one-to-one — we just changed what's doing the answering.

In Search Console Queries & impressions

The searches you appeared for, and how often. You optimise pages against them.

In Citedex Prompts & share of voice

The buyer questions engines were asked, and how much of the answer was you rather than a competitor.

What you do about it Coverage & links, reframed

Blindspots are your coverage report. Citations are your backlink profile — get published on the domains engines already trust.

Citations

The sources AI trusts

Domains cited in this vertical's answers. Get published there and you move the answer — this is the closest thing AI search has to a backlink profile.

Pricing

Founder pricing for the first 20 customers

Lifetime 40% off while we're in beta. Start with a free scan — there's nothing to cancel if it isn't useful.

Starter

$49/mo

1 brand · 25 prompts · 5 competitors · 2 engines · weekly sweeps

Get started

Agency

$399/mo

10 workspaces · white-label PDF reports · client portal links

Get started
Learn more

Frequently asked questions

Which engines do you actually query?

Today: a Llama 3.3 model API (labelled synthetic) and Perplexity's Sonar search API (labelled api). Both are queried directly, and both are named on every metric. We add engines when we can query them properly, not when it would look good on a feature list.

What is an engine "surface", and why does it matter?

A surface is how an answer was obtained. A model answering from its weights, a search-grounded API, and the consumer web app will give you three different answers to the same question — so a single blended "AI visibility score" hides more than it tells you. We never blend them.

How is share of voice calculated?

Each brand's share of all tracked-brand mentions across the prompt set for that week. Brand detection is deterministic — a gazetteer of names, domains and aliases matched against the answer text — so it does not drift when a model changes. We only use an LLM for the softer judgements: sentiment, and whether a mention was a recommendation.

Do you scrape ChatGPT?

No. Everything is queried through official APIs. Scraped consumer interfaces are the web surface, and where we add them, they will be labelled as such rather than quietly mixed into an average.

Can I see the actual answers?

Yes — that's the point. Every raw answer is archived verbatim before anything is extracted from it, and the dashboard links each metric back to the response it came from. If we claim an engine recommended a competitor, you can read the sentence.

What happens after the free scan?

You get a report of what the engines said about you right now, emailed to you and linkable. If you want the week-over-week picture, you can claim the scan into a monitored workspace — no password, just an emailed setup link.

Is your brand in the answers?

Find out in about two minutes. We'll show you the real engine answers, verbatim.

Run a free scan