TL;DR: Choose your first automation by listing the repetitive tasks that take the most hours each week, then scoring them on four things: hours saved, how often the task repeats, how risky a mistake would be and how easy the data is to reach. Start with the task that scores high on value and low on risk. Give it a clear owner, decide where a person approves, and measure hours saved for a month before you expand.
The question I hear most from business owners is not "what can AI do?" It is "where do I start?" Here is the method I use.
Why does the first project matter so much?
Because it decides whether the team trusts automation. A first project that works becomes the reason the next ten get approved. A first project that is too ambitious, too risky or too vague becomes the reason people say "we tried AI, it didn't work."
Step 1: list the repetitive work
Ask each team one question: "What do you do every week that you wish you didn't have to?" You will hear things like:
- Copying leads from WhatsApp into the CRM.
- Chasing customers for documents or payments.
- Building the same report every Monday.
- Answering the same five questions.
- Reminding people about appointments or deadlines.
Write each one down with a rough number of hours per week.
Step 2: score each task
| Factor | Question | Score high when |
|---|---|---|
| Value | How many hours a week does it take? | Several hours, across several people |
| Frequency | How often does it happen? | Daily or many times a day |
| Risk | What happens if it goes wrong once? | A small, fixable mistake |
| Access | Is the data in systems we can connect? | It lives in a CRM, sheet, inbox or calendar |
The best first projects score high on value, frequency and access, and low on risk.
Step 3: decide where a person stays in control
Before building anything, agree on which steps need approval. Anything involving money, prices, contracts or promises to customers should wait for a person at first. I explain the reasoning in People decide, agents execute.
Step 4: give it an owner and a measure
Every automation needs one person who owns it: they review its work, report problems and decide when it is ready to do more. And it needs one number, usually hours saved per week or response time, measured before and after.
Step 5: run it for a month, then expand
A month is enough to see real results and real edge cases. Fix what you find. Then pick the next task from your list. This is how a company builds an AI workforce, one trusted workflow at a time.
What are good first projects?
From what I see across the Gulf and Egypt, these usually work well:
- First reply to new leads on WhatsApp or the website.
- Appointment reminders and confirmations.
- Weekly reports built automatically from the CRM or sheets.
- Invoice and payment reminders.
- Internal requests, like leave, equipment or document requests.
And these are usually poor first projects: anything involving negotiations, large payments, legal decisions or processes that change every week.
Frequently asked questions
How much does a first automation project cost? It depends on the systems involved and the scope. A focused first workflow is usually the most cost-effective way to start.
Do I need to change my software first? Usually not. Good automation works around the tools a team already uses, like WhatsApp, the CRM and spreadsheets.
Should I use automation or AI agents? Use automation for steps that are always the same and agents for steps that need a decision. Many workflows use both. See AI agents vs automation vs chatbots.
How do I measure success? Pick one number before you start, usually hours saved or response time, and compare it after a month.
Who is Amir Diab? I'm the founder of Instanix, a Dubai company that helps businesses in the UAE, the Gulf and Egypt start with automation and AI agents. More about me.
Want this working in your business?
Tell me which task eats your team's week. I'll tell you honestly whether an agent can take it on.