Every few weeks a new AI product promises to "run your business for you". Plenty of owners have tried one, found it impressive for a day, and then gone back to the old way of working. Often the tool wasn't the problem. It was pointed at the wrong job.
This first lesson in the masterclass isn't about any tool. It's about deciding where AI belongs in your business before you spend a rupee on it.
AI is good at three kinds of work
Today's AI systems, the ones behind ChatGPT, Gemini, Claude and Microsoft Copilot, are language and pattern machines. They read, write, summarise, classify and follow instructions. That makes them useful for three broad kinds of work:
- Repeatable work with clear rules. Sending appointment reminders, answering "what are your timings?", copying details from an enquiry form into a sheet, sorting invoices by vendor. The steps are the same every time.
- First drafts. A reply to a complaint, a product description, a quotation cover note, a job post. AI produces a decent draft in seconds, and a person who knows the business fixes the rest.
- Making sense of a pile of information. Summarising 200 customer reviews, pulling the key dates out of a contract, comparing three supplier quotes.
What it is poor at is anything that needs accountability: deciding a price for a difficult client, handling an angry customer who has known you for ten years, choosing which staff member to promote. AI can help you think, but it can't carry the responsibility, and customers can tell when nobody is behind the reply.
Sort your work into three columns
This exercise takes about an hour.
Step 1: list last week's tasks. Yours and your staff's. Be specific: not "marketing" but "posted 3 Instagram stories", "replied to 40 WhatsApp enquiries", "chased 6 unpaid invoices". Your phone's WhatsApp, call log and UPI history are good memory aids.
Step 2: put each task in one of three columns.
| Column | What goes here | Examples |
|---|---|---|
| AI can do it | Same steps every time, low risk if slightly wrong | Booking confirmations, reminders, FAQ replies, data entry |
| AI drafts, a person decides | Needs judgement, but the work is mostly reading or writing | Quotes, review replies, social posts, staff SOPs |
| Stays human | Trust, money decisions, relationships | Negotiation, complaints from regulars, hiring, pricing |
Step 3: put a rough number of hours next to each task. Guesses are fine. You are looking for size, not precision.
Done honestly, the first column is often smaller than the sales pitches suggest, and the second column larger. That's normal. Most of the value from AI in a small business comes from the middle column: work that gets done faster because a draft already exists.
Choose one task, not five
From the first column, pick one task that:
- takes at least two hours a week across the team,
- happens the same way most of the time,
- won't embarrass you badly if it goes wrong once.
That is your first project. A salon might pick appointment reminders. A CA firm might pick the monthly "please send your bank statements" chase. A distributor might pick turning WhatsApp orders into a clean sheet.
Starting with one task matters for a practical reason: you'll learn how AI behaves with your customers and your data, and that lesson transfers to everything else. Owners who start with five projects usually finish none.
Where the hype goes wrong
Three claims you'll hear often, and what they leave out:
- "Replace your staff with AI." In most small businesses, AI removes tasks, not roles. The receptionist is still needed, and spends less time answering the same question and more time with walk-ins.
- "Set it and forget it." Anything customer-facing needs someone checking its output, especially in the first months. We cover this in detail in human in the loop.
- "Works for any business." The same tool can save a clinic hours and do nothing for a factory. Your three columns tell you which, before any vendor does.
What this costs to try
You don't need a budget to begin. The free tiers of the main assistants are enough to test drafting and summarising, and India-specific plans start low: ChatGPT Go, for instance, costs ₹399 a month. Automating a customer-facing task such as WhatsApp replies costs more, and we break those costs down in how much AI costs a small business in India.
Before paying anyone, read is ChatGPT safe for business data?. What you paste into a free tool matters.
Your checklist for this lesson
- List every task from last week, yours and your staff's
- Sort each into: AI can do it / AI drafts / stays human
- Add rough weekly hours to each task
- Circle one repeatable task worth 2+ hours a week
- Write down what "working" would look like for that task (fewer missed replies? hours saved?)
Keep that sheet. Every later lesson in this masterclass builds on it. If you'd like a second opinion on where AI fits in your business, our AI consulting starts with exactly this exercise.
Next in the masterclass: Part 2, your first week with an AI assistant.
