Who does AI actually
recommend in Dubai?

The boards: same market, two vocabularies

Each series counts the brands named in completed answers. The mix of software, directories, and agencies below is not noise — it is literally what the engines name, and the difference between the two columns is the finding.

How to legit-check any agency in Dubai — including us

“Not a scam” is a real query in this market — we measured it. The engines resolve that worry with third-party evidence, and you should demand the same. Five checks, applicable to us word for word:

1. 01

   Ask for dated measurements, not promises.

   A legitimate measurement claim names the date, the engines, the question set, and the denominator. Every number on this page carries all four. If a pitch can't produce that, it's a forecast wearing a lab coat.

2. 02

   Check the third-party record.

   The engines lean on directories and reviews (Clutch appeared in 46.4% of street answers). Look the vendor up where the engines look. We publish our own directory and citation record rather than asking you to take our word.

3. 03

   Ask which engines they measure — and whether web search is on.

   Answers change materially with retrieval on. A “ChatGPT ranking” measured without web search is a different instrument than the one your customers use.

4. 04

   Demand the price in writing, in AED, with scope.

   The table above is the market's published range. Our own prices are public: $99 / $399 / $1,500 per month (about AED 365 / 1,465 / 5,510 at the dirham's 3.6725 USD peg), billed in USD, scope on the pricing page.

5. 05

   Run their own test on them.

   Ask the assistants about the vendor by name, then ask the category questions without their name. We publish our own split every week — branded 100%, Dubai discovery 0% as of this study — because a measurement company that hides its own reading has told you everything.

Methodology, limits stated

The same instrument as our published methodology: frozen query sets, five engines, repeat sampling on high-intent queries, brands counted from completed answers. What a reading like this can and cannot claim:

• The market series (run 55bc15a5) completed 126 of 140 engine-query cells; the run stopped at its spend ceiling, and all rates are computed over the 126 completed cells. The street-vernacular series (run e6265f99) completed 84 of 85 cells.
• Queries were locale-conditioned in the query text only (Dubai/UAE qualifiers); API calls originated from US infrastructure with no engine-side location setting. A location-parameter instrument is a future series version and will be labeled separately.
• Google-surface readings requested a Dubai locale via a third-party SERP provider; 12 of 28 locale-conditioned requests failed and are excluded from denominators rather than counted as non-mentions.
• Gemini declined to name providers on a portion of medical/aesthetic and legal queries; refusals are recorded as refusals and excluded from that engine's mention-rate denominators, not counted as non-mentions.
• These are API-surface readings. They measure what the engines' APIs return to locale-conditioned queries, pending consumer-surface receipts from devices in the UAE; consumer apps can differ in retrieval, safety behavior, and localization.

A single reading is a dated snapshot, not a standing ranking. When the Dubai panel re-runs, new points append to these labeled series; published readings are never recomputed or re-dated.