Buyer guide · Artificial intelligence

AI companies in Saudi Arabia: how to tell them apart and choose one

"AI company" covers at least five different kinds of business in the Kingdom, from compute providers to product companies to integrators. Decide which kind your problem needs first, and the shortlist gets much shorter. This guide sets out the five kinds, how to match one to your problem, the questions that are specific to Saudi Arabia, and where Interkey fits.

Published November 2, 2026 9 min read

Published by Interkey. This guide does not rank or name other companies.

Put your AI question to the team See the eight questions

In short

There is no universal list. Directories of AI companies in Saudi Arabia mix global vendors, startups, agencies and integrators, and seldom say how they ordered them.
Start from the kind of company: infrastructure and model providers, AI product companies, AI services practices, system integrators, and implementers of global platforms.
Then ask the Saudi questions: where data and inference run, how the system was tested on Arabic data, how it maps to SDAIA's AI Ethics Principles, and who operates it after handover.
Where Interkey sits: a Riyadh company with its own computer-vision product (SquintPRO), an enterprise AI practice and integration work. It is not a compute or foundation-model provider.

Five kinds of AI company, and what each one actually sells

Most shortlists go wrong at the first step: a GPU provider, a video-analytics product and a data-science consultancy are compared as if they were alternatives. They are not. Each sells something different and is accountable for something different.

Five kinds of AI company and how to recognise each
Kind of company	What you buy	It fits when	First question to ask	Warning sign
Infrastructure and model providers	Compute capacity, hosting, data-centre space, or access to large language models	Your own team will build on raw capacity or a model API, and the decision is about where it runs and what it costs	Where, physically, do our data and inference run, and under which contract?	A capacity offer presented as the answer to a business problem
AI product companies	A packaged product that solves one class of problem, licensed and configured	Your problem matches what the product already does, such as detecting safety events on existing cameras	What does the product do today, without custom development, and how is it licensed?	A roadmap answer to a question about current features
AI services and data-science practices	A model or pipeline built on your data	The problem is specific to your organisation and the data to solve it exists	What would make you advise us not to build this?	An accuracy figure quoted before anyone has seen your data
System integrators	Getting AI output into the systems that act on it: ERP, CRM, ticketing, cameras, workflows	The model exists or is bought; the hard part is daily operation	Who owns each interface, and what happens when one side changes?	No named owner for each interface
Implementers of global platforms	Deployment, configuration and support of another vendor's AI-enabled platform	You have chosen the platform and need it run properly in the Kingdom	Which parts are the vendor's and which are yours, and how is each priced?	The implementer and the vendor blurred into one "we"

Many companies are more than one kind. What matters is knowing which kind you are buying in this contract, because that decides who is accountable when the system misbehaves.

Start from the problem, then choose the kind

The same budget line can hide very different work. Use the situation you are in to decide which kind of company to call first.

Which kind of AI company to call first, by situation
Your situation	Call first	Why
"We have many cameras and a safety or operations question"	An AI product company with a video-analytics product	Detection, search and alerting already exist; the work is configuration and a pilot on your own sites
"We want to predict failures, demand or fraud from our records"	An AI services practice	The outcome depends on your historical data, so a feasibility and data audit comes before any model
"We want an assistant that answers from our documents"	A services practice, with an integrator	The difficult part is the document collection and where the model may run, not the chat window
"We need GPU capacity or a hosted model for our own team"	An infrastructure or model provider	You are buying capacity, so compare location, contract terms and cost at your volume
"We have chosen a platform and need it live"	An implementer of that platform	You need people who run it every day, in the Kingdom
"The decision can be written down as a rule"	Probably no AI company	A rules engine is usually cheaper, easier to audit and does not drift

For what an AI request usually turns out to be once the data is examined, see Enterprise AI in Saudi Arabia.

Four questions that are specific to Saudi Arabia

A supplier can be strong in other markets and still be a poor fit here. These four questions separate the two.

1. Where does the data go, and where does inference run?

Ask for an answer per deployment option, not one answer for the whole system. A hosted model API means your inputs leave your estate; software you run on your own infrastructure keeps them in. For some classes of data either is acceptable; for others only one is. Your data classification should settle this before the architecture is drawn.

2. How was it tested on Arabic data?

Documents, tickets, call transcripts and customer messages in the Kingdom contain Arabic, mixed Arabic and English in the same field, and transliterated names. General-purpose models handle these unevenly. Ask to see an evaluation on samples of your own Arabic data, not a benchmark from another market.

3. How does the system map to SDAIA's AI Ethics Principles?

The Saudi Data and AI Authority (SDAIA) published its AI Ethics Principles in September 2023 for anyone who designs, develops, deploys, implements or uses AI systems in the Kingdom. They set seven principles: fairness; privacy and security; humanity; social and environmental benefits; reliability and safety; transparency and explainability; and accountability and responsibility. They also describe four levels of risk, from little or no risk to unacceptable risk.

A serious supplier can tell you which risk level it would assign your use case and why, and how its design supports each principle that applies. Buying from a supplier does not remove your share of the accountability: the organisation that deploys the system remains answerable for how it is used.

4. Who runs it after go-live?

Building a model is a project; running it for years is a capability. Ask who monitors it, what triggers retraining, who is authorised to switch it off, and whether that team works in the Kingdom, on your working week.

How to read a "top AI companies" list

Lists are a reasonable place to collect names and a poor place to choose between them. Saudi search results for this topic, checked in November 2026, put global technology vendors, startups, development agencies and consultancies in the same lists, and rarely explained how the order was decided. Before relying on a list, check three things:

Does it say how companies were selected and ordered? If not, treat it as a directory, not a ranking.
Does it separate the five kinds? A list that ranks a model provider against an integrator is comparing different purchases.
Does it show where each company delivers from? Local delivery affects support hours, Arabic and where your data goes.

Eight questions for the first call with any AI company

Use these to bring a long list down to three. They apply to Interkey too. For the requirements to write into a formal RFP, use the procurement list on Enterprise AI in Saudi Arabia.

Which kind of AI company are you in this contract: product, services, integrator, platform implementer or infrastructure?
What do you own, and what belongs to a partner? Name the partner for each component.
Where does our data land, and where does inference run, for each deployment option you offer?
Can you show an evaluation on our own data, including Arabic text if our use case has it?
Which SDAIA risk level would you assign this use case, and what does that change in your design?
What does the system do when it is unsure or unavailable, and which human process takes over?
Who operates the system after handover, from where, and during which hours?
Can you arrange a reference in our sector, with that organisation's agreement?

RelatedEnterprise AI in Saudi Arabia

Where Interkey fits, on facts it can support

Interkey is a Saudi enterprise technology company, founded in Riyadh in 1999. In the five-kind map it is three of the kinds, and it is clear about the one it is not.

An AI product company: SquintPRO

SquintPRO is Interkey's own computer-vision product. It reads the CCTV feeds a site already has. It runs either on-premise or in the cloud, with 3, 6 or 12 months of footage retention in the cloud option, and it is priced per use case. It comes with 35 trained libraries, 12 core use-case categories and 73 configurable metrics and alerts. A new event can be defined with a sentence or a reference image, footage can be searched in plain words or with an image, and a vehicle can be followed across cameras and shown on a map. SquintPRO does not use face recognition. It is in live use in energy, ports, smart-city projects and Interkey's own factory.

SquintPRO

An AI services practice

The enterprise AI practice works in four stages: feasibility and data audit, modelling and evaluation, production engineering, and operation and retraining. Stopping after any stage is a legitimate outcome, including a documented "do not build" after the data audit.

Enterprise AI in Saudi Arabia

An integrator and operator

Interkey connects model output to the systems that act on it, and treats a production model like any other service: instrumented, alerted on and visible to the team on call. Interkey is an official Datadog partner.

Observability practice

Not an infrastructure or foundation-model provider

If you need GPU capacity or a large language model as a service, start with a provider of those. Interkey can be the integrator around it.

Delivery from Riyadh on the Saudi working week. Certified to ISO 9001:2015 and ISO/IEC 27001:2022. No named-customer case study is published yet; a reference in your sector is arranged on request, with the other organisation's agreement. Case studies and deployment patterns

What buyers ask us

Which are the top AI companies in Saudi Arabia?

It depends on what you are buying. Interkey does not publish a ranking, because a ranking that mixes infrastructure providers, product companies and consultancies compares different purchases. Decide which of the five kinds you need, then compare three companies of that kind using the eight questions above.

Should we choose a Saudi AI company or a global one?

Choose on the answers rather than the headquarters. Global platforms can be delivered in the Kingdom through local implementers (Interkey implements several), so the practical questions are where data and inference run, how Arabic was tested, and who operates the system here after go-live.

Do SDAIA's AI Ethics Principles apply to private companies?

SDAIA wrote them for anyone who designs, develops, deploys, implements or uses AI systems in the Kingdom, public or private. How they apply to a specific system is a question for your own governance and legal teams; a supplier can show how its design supports them, but cannot certify you against them.

Is a proof of concept worth doing?

Yes, when it is designed to answer the question that decides the purchase. Agree the success criteria, the data and the point at which you stop before it starts. For camera-based safety monitoring, this is set out in Scoping a pilot that proves something.

Continue evaluating

Next step

Bring the problem, and we will say which kind of company it needs

Describe what you are trying to decide or improve. An engineer from the Riyadh team replies with the kind of supplier that fits, including when that is not Interkey.

First name Last name

Company Email Company

Phone number (optional) What are you working on? (optional)

Required field

The Riyadh team reviews your request and a specialist replies on Saudi working days, in Arabic or English. Your details are used only to reply to you. Privacy Notice

Published by Interkey. Last updated 2026-11-02. Interkey is registered in Riyadh, Saudi Arabia under commercial registration 1010156897.