Do not burn tokens watching. Wake your agent only when it matters.
AI Brand Watch

Know the day ChatGPT stops recommending you.

AI Brand Watch asks the questions that define your category to ChatGPT, Claude, Perplexity and Gemini — on your schedule — and alerts you the moment your mention rate changes. It measures the answers, mentions, citations, position and competitors the models return for those queries.

Open Brand Watch on Apify → See a sample alert

Pay per check · no subscription

How it works

Three steps. Then silence — until something changes.

01 · Just a brand name

No query list required. Desvela discovers the 5–30 questions buyers actually ask AI in your category — plus your competitors and domain — and freezes them as a prompt set. The first run is an audit: it establishes your baseline, silently.

02 · We sample the engines

Each question is asked several times per engine, because AI answers vary. You get mention rate — the % of answers that mention you — plus competitive share of voice, owned vs earned citations, and the sources each engine cites. Every scheduled run measures the same frozen questions, so the number is comparable week to week.

03 · You only hear about changes

You appeared, disappeared or shifted 25+ points — with the reason, the before/after, and the new source or competitor that explains the move. Delivered as a signed webhook your systems can verify and to your Apify dataset; pipe it on to Slack, email or Zapier through Apify's integrations.

What an alert looks like

The receipt, not the dashboard.

No login, no charts to interpret. When something changes, you get the event: what changed, before and after, and the source that changed the engine's mind — so your next move is obvious.

CHANGE EVENT
№ 0117 · HMAC signed
Acme CRM disappeared from Perplexity's answer.
Query"best CRM for small business in Spain" · Perplexity
Mention rate67% → 0% · new source cited: reddit.com/r/…
Delivered07 · 07 · 2026 — webhook · 210 ms
Real numbers, our own brands

0% is the most valuable number we've shipped.

Before charging anyone, we ran Brand Watch on two businesses we own. Full disclosure: these are our brands — that's why we can show you the raw numbers.

Case 1 · Digital time capsules

Mention rate: 0% across ChatGPT, Claude, Perplexity and Gemini — completely invisible.

But the run returned the exact sources the engines cite instead: YouTube dominates, and Arweave, IPFS and OpenTimestamps own the category. That list is the GEO to-do list — one run, ~$5.

Case 2 · Photography backdrops store

Brand mentions: 0%. Yet its URL is cited by Perplexity in 4 of 5 queries.

Brand mention and source citation are different metrics — we measure both. And each engine cites a different ecosystem: Gemini favors photo blogs, Perplexity cites the store and Reddit. Competitor benchmark included (Manfrotto: 100% on the head query).

What the watch caught next

On the purchase-intent query — "where to buy" — the store appeared in 1 of 3 samples on Perplexity.

A single check would have called that yes or no at random. Sampling turns it into a rate worth tracking — the weekly watch now guards exactly that number, and alerts only if it moves.

Pricing

Pay per check. Nothing else.

EventPrice
Actor start$0.02
Brand query checked (per query × engine, all samples included)$0.08
Prompt discovery (zero-config only, when a prompt set is generated)$0.04
GEO recommendations (opt-in, only charged when delivered)$0.04
Sentiment (opt-in, per measured cell)$0.04
Change alertsfree

Typical setup — 20 queries × 2 engines, weekly: ≈ $14/month (the SaaS equivalent charges $99–300/month flat).

Turn on includeRecommendations and get 3–5 concrete GEO actions from that run's data — which sources each engine cites, where competitors win — grounded in your actual results, not generic advice. Turn on includeSentiment and each measured cell also carries a positive/neutral/negative label — $0.04 per cell, never triggers alerts.

Honest measurement

We refuse to sell you noise.

AI answers change every time. How is this reliable?
That's exactly why we sample: each question is asked N times per engine and reported as a mention rate — the % of answers that mention you — not as a single screenshot. Alerts only fire past defined thresholds.
Do I need to know the exact questions buyers ask?
No. With just a brand name, Desvela discovers the category questions and competitors for you and freezes them as a prompt set. The first run audits and establishes your baseline; every later run measures the same frozen questions, so the diff means something. You can also pass your own list — both work, and a prompt set never changes silently under a watch.
What's competitive share of voice?
Mention rate is your brand's presence in the answers. Competitive share of voice measures you against the competitors you're actually watching — your share of the answers where you or a competitor appears. With your domain and competitor domains, you also get owned vs earned citations: how often the engine links your site versus anyone else's.
Is this what the real ChatGPT says?
We call the engines' APIs with web search enabled. API answers correlate with, but are not pixel-identical to, each consumer product. We always measure the same way, so the diffs — the part you pay for — stay valid.
Do I need API keys?
No. Checks run on ours; the per-check price includes it.
Can my AI agent consume this instead of me?
Yes — that's the point. See For agents.
Run a brand audit →