By Fritill team ·

*Last updated: 19 November 2026 · By the Fritill team*
Short answer: being AI-ready is less about choosing a model and more about whether your data, processes and people can support a first, measurable use of AI. Work through ten questions, each scored 0 to 2, across those three layers. Your lowest-scoring layer is the first thing to fix. In our experience, the early wins in many businesses come from data and integration work, not from a new chatbot.
As more people ask AI tools to compare options for them, and some start letting those tools book and buy, and as more internal work is drafted by tools rather than typed from scratch, the gap will widen between companies whose information can be trusted and companies whose information still lives in chats and spreadsheets. The question for businesses in the UAE, Saudi Arabia and Egypt is not "should we try AI?" It is "what would have to be true before a first project actually holds?"
This is Fritill's own self-check, not a published industry index. It takes about ten minutes. Answer for the business as it is today, not as it is planned.
What does "AI-ready" actually mean?
It means you can trust the data a tool would use, you can describe the work you want to change, and someone owns what happens next. A business can buy a licence tomorrow and still not be ready. Readiness is the ground under the first project, not the project itself.
A useful first project has a clear problem, a way to tell whether it worked, and data that would not embarrass you if a report were built on it. If those are missing, the tool will mostly automate the mess.
We argued the same order of thinking in your next customer might not be human: agents and people alike get stuck when prices, availability and policies cannot be read cleanly. It also sits behind our predictions for 2027: code is getting cheaper, and the scarce work is deciding what to change and reviewing the result.
Which three layers should you check first?
Check data, then processes and systems, then people and direction. AI sits on top of those layers. If the data is scattered, the process lives in one person's head, or nobody owns the decision, a new tool will not fix that.
Data. Do your customers, orders, products and prices live in a system you can query? Is that data complete enough that you would trust a report built on it? Do you know who can see it, and how personal data is handled under the law that applies to you in the UAE, Saudi Arabia or Egypt? *(That last point is a prompt to check, not legal advice.)*
Processes and systems. Are your most important workflows (enquiries, quoting, invoicing, or whatever yours actually are) written down and done the same way each time? Do tools pass information to each other, or does someone retype it? Could an outside reader, including an AI agent, find your prices, availability and policies in a structured form, not only in PDFs, images or WhatsApp? Do your names, policies and content exist in Arabic written for Arabic readers, as well as English, rather than only as a machine translation?
People and direction. Can you name one or two specific problems where AI or automation should save time or money in the next 6 to 12 months? Does someone senior own technology decisions, even part-time? When you try a new tool, do you agree in advance how you will measure it, and does the team actually use it?
The lowest-scoring layer is where to start, even if a more visible AI idea is tempting.
How do you run the 10-question self-check?
Answer each question for the business today: 0 = not at all, 1 = partly, 2 = yes, and we could show you. Add the ten scores for a total out of 20. Copy the list into a document or print this page if you want to score it with your team.
Data
- Our core business data (customers, orders, products, prices) lives in a system, not in scattered spreadsheets, PDFs and chats.
- That data is reasonably complete and correct. We would trust a report built on it.
- We know who can access which data, and personal data is handled under the data protection law that applies to us (UAE, Saudi Arabia or Egypt).
Processes and systems
- Our three most important processes (for example enquiries, quoting, invoicing) are written down and done the same way every time.
- Our main tools are connected. Information doesn't get retyped from one system into another.
- Our prices, availability and policies are published in a structured, readable form (on our website or in a feed), not only in PDFs, images or WhatsApp.
- Our product names, policies and content exist in native Arabic as well as English, not only as machine translation.
People and direction
- We can name one or two specific problems where AI or automation would save time or money in the next 6 to 12 months.
- Someone senior owns technology decisions (architecture, vendors, security), even if only part-time.
- When we try a new tool, we agree up front how we'll measure whether it worked, and the team actually adopts it.
What does your score mean?
The number is a starting point, not a grade. Read it alongside your lowest-scoring section. A high total with very weak data is not ready for a wide rollout. A modest total with one clear, measurable problem can be ready for a small pilot.
| Score | Where you are | Start here |
|---|---|---|
| 0 to 7: Foundations first | AI would mostly automate the mess. | Pick one process. Write it down, move its data into one system, and connect it to one other tool. Questions 1, 4 and 5. |
| 8 to 14: Ready for a first pilot | You have enough structure for one small, measurable AI or automation win. | Choose the problem from question 8 that's closest to revenue. Define what "worked" means before you build anything. |
| 15 to 20: Ready to scale | Your data and processes can carry AI beyond a pilot. | Decide who owns AI decisions, budget and risk, and look at where agents and customers will read your business (question 6). |
These bands are ours, for this self-check. They are not a certified score and they are not a promise about what a later project will return.
What should you do first if the foundations are weak?
Don't start with a chatbot. Pick one process that already makes or loses money, write down how it works today, put its records in one place, and only then decide what to automate.
A written process is a short, shared description: what starts the work, what "done" looks like, who can approve an exception, and where the data sits. Once that exists, connecting two tools is usually worth more than adding a new interface on top of the old retyping.
If personal or payment data would be involved, run a basic access and security pass before a tool touches it. Our 12-point check for AI-built apps is a practical starting point when the first version is built with AI tools.
What does a first pilot look like?
A first pilot is one problem, one measure, one owner, and a date to decide whether to keep it, change it or stop. It is not a company-wide AI programme, and it is not a tool chosen before the problem is named.
Write four lines before anyone builds: the problem, the user, the measure, and what you will leave out. Then run it long enough to see a real cycle of work, not a demo. If you cannot write the four lines, you are still in the foundations stage, and that is a better place to spend the next month than a licence you won't be able to judge.
In a few years, the businesses that are easiest to work with, for people and for AI agents, will be the ones whose prices, availability and policies can be read, whose records can be trusted, and whose team can tell whether a change worked. That is the point of getting ready, more than the tool itself.
How we'd approach this
We'd start with the score, then spend most of the first conversation on the lowest-scoring layer and on the problem from question 8 that sits closest to revenue. We would not begin by choosing a model.
If you want to talk that through, bring your answers to a free 30-minute intro call. If you'd rather have a structured look, Fritill's AI & Automation Assessment is a two to three week assessment of up to 3 processes: a prioritised list of AI and automation opportunities with effort and expected value, and a recommended first step.
If you would rather keep reading: for businesses, or what we do.
Fritill is a technology venture studio and digital transformation partner. We build and run our own products, and we help founders and businesses in the UAE, Saudi Arabia and Egypt get the ground ready for a first project in the same order we use ourselves: data and processes first, then the tool.
---
*This self-check is Fritill's own framework (version 1, November 2026). It is not a certified or third-party index, and the data protection prompts are not legal advice.*
Questions
What is an AI readiness assessment?
A structured way to see whether your data, processes and people can support a first, measurable use of AI, and where to start. This self-check is ten questions, each scored 0 to 2, with a total out of 20. It is Fritill's own framework, not a published industry index.
How long does the self-check take?
About ten minutes, if you answer for the business as it is today. The useful part is being honest about the lowest-scoring layer, not finishing quickly.
Does a high score mean we should buy an AI tool?
No. A high score means the ground is firm enough for a first pilot, or to go beyond one. You still need a named problem, a measure and an owner.
What if our score is low?
Start with foundations: one process, its data in one system, and a link to one other tool. A low score is not a failure. It is a map of the work that will make any later AI project easier to judge.
Is this the same as Fritill's AI & Automation Assessment?
No. This self-check is free and you can run it yourself. The Assessment is a paid, fixed-scope engagement described on the [pricing page](/en/pricing#ai-assessment): up to 3 processes, a prioritised list of AI and automation opportunities with effort and expected value, and a recommended first step. Start with the self-check.
Do we need Arabic content to be AI-ready?
If you serve Arabic-speaking customers, yes, in the sense that names, policies and offers should exist in Arabic written for those readers. Machine-translated pages are harder for people, and for agents, to trust. If you only serve one language today, say so honestly on question 7 and plan the second language when it becomes a real need.