Fazley Rabbi"Everyone says we should use AI, but I do not know where to start, and I do not want to spend money on something my staff will not use." The first automation project sets the tone. If it works, your team trusts the next one. If it is too big, unclear or risky, it gets abandoned and everyone decides AI "does not work for us". The answer is to choose the first project deliberately, and keep it small.
Signs of a good first project
Look for a task that is:
- Repetitive and frequent. Done daily or many times a week, the same way each time.
- Rule-based. You can explain the steps to a new staff member in a few minutes.
- Already digital, or close to it. The information is in a sheet, an email, a form or a message, not only in someone's head.
- Costly in time or errors. Someone spends real hours on it, or mistakes cause rework or unhappy customers.
- Low risk if something goes wrong. A mistake is easy to spot and fix before it reaches a customer or touches money.
- Owned by someone. One person cares about the result and will tell you if it breaks.
Typical candidates in small businesses include a daily sales summary, order confirmation messages, copying invoice data into a sheet, or appointment reminders.
Signs of a poor first project
- A chatbot that answers customers on its own, including prices and promises, with nobody reviewing.
- Anything that moves money or approves refunds without a person.
- A process each staff member does differently, or that changes every month.
- A vague goal such as "use AI across the company" with no specific task.
A step-by-step way to choose
- List tasks for one week. Ask each staff member to note the repetitive work they did, how often and how long it took.
- Score each task from one to five on frequency, time spent, cost of errors and risk. Strike out anything high-risk.
- Pick one, usually the task with the best mix of frequent, time-consuming and low-risk.
- Write down the current steps exactly, including the exceptions ("if the customer is outside Dhaka, we...").
- Fix the process before automating it. Automating a messy process only makes the mess faster.
- Decide what success looks like in your own measures: hours saved each week, fewer errors, faster replies. Note today's figure so you can compare honestly.
- Build a small version and run it alongside the manual process for a couple of weeks. Compare results before switching over.
- Review, then expand to the next task on your list.
Expensive mistakes to avoid
- Buying software before defining the problem. Tools are easy to buy and hard to adopt.
- Too big a scope. One process done well beats five half-finished.
- No plan for failures. Decide what happens if the internet is down, the power goes out or a tool's service changes: who does the task by hand, and how you find out it has stopped.
- Not owning your accounts and data. Logins, automations and data should be in your business's name, not only a freelancer's.
- No human review for sensitive outputs. Anything about prices, refunds, medical or legal matters is checked by a person before it is sent.
- Leaving staff out. The people doing the task know the exceptions. Involve them early, and be clear about how the change affects their work.
Your part, and where an expert helps
You are best placed to list the tasks, score them, write down the current steps and judge the results. An expert helps choose suitable tools, connect your systems, build it so it runs reliably, handle security and access, and hand over clear documentation so you are not dependent on one person. When I start with a business, I usually spend the first conversation on the task list rather than on tools. For a second opinion on your list, see my AI and workflow automation service, or look at a common first project such as a daily sales report.
The smallest useful first step
Start the one-week task list tomorrow. A week from now you will have a page of repetitive work, and one or two tasks will clearly stand out. That page is the brief for your first project. If you would like help reading it, get in touch.