What a UAE Business Should Expect From an AI Consulting Firm

If you are the general manager or COO of a trading, contracting or real estate business in Dubai, Abu Dhabi or Sharjah, you have more AI advice on offer than you can use. The harder question is what to expect from an AI consulting firm once you engage one, so that you can tell a useful engagement from an expensive one in the first fortnight.

This post sets out what a UAE business should reasonably expect: how the first project should be scoped, the data residency and Personal Data Protection Law questions to put on the table before any file is shared, how Arabic and English should be handled, and what you should own at the end. For a checklist on evaluating and comparing firms, see our guide on evaluating AI consulting companies in India; the evaluation logic is the same, so we will not repeat it here.

What an AI consulting firm should do first: a scoped project, not a roadmap deck

The UAE has a national AI strategy with objectives that include adopting AI across customer services and ensuring strong governance. That ambition at the federal level has encouraged a lot of boardroom conversation about "AI transformation". For a mid-sized operator, that conversation is a distraction until one workflow is running.

An AI consulting firm earning its fee will, within the first few weeks, hand you a one-page specification for a first project. It should name:

• The workflow (for example, turning supplier quotations into a priced customer quote, or pre-checking shipping documents before they reach your clearing agent).
• The inputs it will read and the systems it will write to.
• The exception rule: what the system does when it is unsure.
• The person who reviews the output, and at what point.
• The measure you will use to judge it after a set number of weeks.

Ask for this document by name. If what arrives instead is a thirty-slide maturity assessment, you are paying for the firm's sales cycle.

A good firm will also tell you when AI is unnecessary. Many first wins in UAE back offices are plain workflow rules: a reminder when a quotation is unanswered, a check that a delivery note matches a purchase order, a status update that goes out when a milestone is logged. AI earns its place where the input is free text, such as reading an enquiry in mixed Arabic and English and working out what is being asked. Our post on AI adoption in India and the UAE covers where businesses in both markets tend to start.

Data residency and the UAE PDPL: questions to ask before you share a file

The UAE's Personal Data Protection Law, Federal Decree-Law No. 45 of 2021, has been in force since 2 January 2022. It applies to processing of personal data inside the country and, in defined cases, outside it. Processing without the data owner's consent is generally prohibited except where the law allows it, individuals can ask for corrections and for processing to stop, and the law sets requirements for transferring personal data across borders. The UAE Data Office is the federal regulator under the law.

Most AI tooling involves sending text to a model hosted somewhere. That makes residency a first-meeting question, not a legal-review afterthought. Put these to the firm in writing:

1. Where will each copy of our data live? Your own systems, the firm's working environment, any model provider, any messaging platform. Ask for a simple diagram.
2. What is the basis for each transfer outside the UAE, if any? The firm should be able to explain its approach and defer to your legal adviser on the final call.
3. Will any provider use our data to train its models? Major providers' business and API terms generally say no by default; OpenAI's enterprise privacy page, for instance, states that business data is not used for training. Ask for the equivalent statement from every provider in the proposed design.
4. Which regime applies to our entity? If you are licensed in a financial free zone, ask the firm and your adviser which data protection rules govern you.
5. How is deletion handled at the end? Including test data, logs and any copies made during development.

A firm that treats these as routine has worked in the UAE before. One that treats them as obstacles has not.

Arabic and English in the same workflow

Your customers and suppliers write in Arabic, English, and often both in one message. A system that performs well in English demos and fails on a WhatsApp message that switches language halfway through is worse than no system, because your team will stop trusting it.

Expect the AI consulting firm to:

• Build a test set from your real messages (anonymised), including transliterated Arabic in Latin script and mixed-language sentences.
• Show you how the system behaves on the cases it cannot classify, and confirm that those go to a person.
• Handle Arabic in outgoing messages where your customers expect it, with a native speaker reviewing templates before they go live.
• Record which language each contact prefers, so that follow-ups do not flip between the two.

Ask to see the test results, not a description of them.

What the firm should hand over

By the end of a first engagement you should own everything needed to carry on without the firm. The table below is a reasonable minimum.

Confirm intellectual property ownership in the contract before work starts. The post on automation provider exit drills is a useful rehearsal for what happens if the firm vanishes.

Where AI sits in a UAE back office: a short map

To make the first project concrete, here are the kinds of workflow a Dubai operator typically puts forward, grouped by what the firm should do with each.

• Plain rules first: quotation reminders, delivery note versus purchase order checks, milestone status updates, document expiry alerts for trade licences and insurance.
• AI where text is involved: reading enquiries and tagging what is asked, drafting a first-pass quote from a specification for a person to approve, summarising a long email thread into a task.
• AI with extra caution: anything that contacts customers directly, commits a price, or touches payment data. These need a human approval step in the design, not as an afterthought.

For an illustrative setup, consider a mid-sized UAE contracting firm whose enquiries arrive by email and WhatsApp. A quoting assistant could collect the scope, flag missing drawings and prepare a draft from an approved rate library. An estimator would approve exclusions and prices before anything reaches the customer. The first pilot would test that handoff rather than promise a particular saving.

What drives the cost of a first project in the UAE

Do not expect a rate card, and be suspicious of one that arrives before discovery. The cost of a first project with an AI consulting firm moves on a handful of factors, and a proposal that separates them is easier to compare with the next one:

• Scope. A fixed-scope pilot on one workflow costs less and is easier to judge than open-ended time and materials. Prefer fixed scope the first time.
• Systems. Every additional system the workflow reads from or writes to adds integration work, and older or on-premise systems add more.
• Languages. Bilingual Arabic and English handling adds testing and template review time; budget for it rather than discovering it in month two.
• Data preparation. If your records are spread across spreadsheets and inboxes, part of the first project is consolidating them, and that should be priced visibly.
• Run costs after go-live. Model usage, messaging fees and hosting are usually small per transaction but add up at volume. Ask for a monthly estimate at your expected volume in AED, and for the official pricing pages behind it.

If a proposal cannot show these lines separately, ask for it to be rewritten before you compare it with anyone else's.

How to decide: a short checklist

Before you sign with any AI consulting firm, confirm you can answer yes to each line:

1. The first project is written down on one page with a measure attached.
2. You have a diagram of where your data goes and the basis for each transfer.
3. Arabic and English handling has been tested on your own messages.
4. There is a named human reviewer and a rule for exceptions.
5. You will own the documentation, accounts and build at the end.
6. There is a named owner on your side.

Two or more gaps means the scope is not ready. Fix the scope before fixing the price.

Where to go from here

List the three processes that cause the most rework in your operation this month and bring that list, not a technology wish list, to the first meeting. Our AI consulting page explains how we scope a first project, our Dubai AI agency page covers delivery in the UAE, and you can talk to us about your workflow if you want a second opinion on a proposal.