How to start a business with AI

01

Start with a rough idea, not a business plan

Most people stall here because they think the idea has to arrive finished. It doesn't. A sentence about a problem you've noticed is enough to work with.

AI is genuinely good at this part: you describe what you've been thinking about in ordinary words, and it asks the questions that turn a hunch into something specific. Who is it for? What do they do today instead? What would make them switch?

• Write the idea the way you'd say it to a friend.
• Name one person who would use it, not a whole market.
• Don't name the business yet. That decision gets easier later.

02

Check whether it's worth doing

Before you build anything, work out who it's for, what problem it solves, how it could make money and what could go wrong. This is the step people skip, and it's the reason most first attempts quietly disappear.

You don't need a research budget. You need honest answers to a handful of questions, and the cheapest possible test of the riskiest assumption: usually whether anyone actually wants this.

• Write down the one thing that has to be true for this to work.
• Find the cheapest way to test it: ten conversations, a waiting list, a single post.
• Decide in advance what result would make you stop.

03

Decide what the first version actually is

The first version is not the business. It's the smallest thing that lets a real person get the result you're promising, and lets you learn whether they care.

For most ideas that's a simple site or app: explain the offer, let someone sign up or enquire, and handle the rest by hand at first. Doing things manually early is not cheating. It's how you find out what to automate.

• One clear promise on the page.
• One action you want a visitor to take.
• Everything else can wait for version two.

04

Build it with AI

This is where AI tools earn their reputation. You describe what you want in plain English and get a working website or app back, without writing code.

The quality of what you get back depends almost entirely on the quality of the instructions you give. A vague request produces a generic result. A brief that names your customer, your offer, the pages you need and the action you want produces something you can actually use.

Write the brief first, then build. Thinking and building are different jobs, and doing them in the wrong order is the most common way people waste weeks.

05

Put it in front of real people

Launching is not an event. It's showing the thing to ten people who match your customer, watching what they do, and fixing what confused them.

Look for the moments people hesitate: the headline they don't understand, the price they don't expect, the button they don't press. Those are your next three jobs.

• Show it to people who could genuinely buy, not just friends.
• Ask what they thought it was for before you explain.
• Change one thing at a time so you know what worked.

06

Get the first sale, then improve

A business starts at the first paid customer, not the first launch. Everything before that is preparation.

After that it becomes a loop: decide the next move, build it, see how it lands, learn something, improve. AI shortens every step in that loop, but it doesn't replace the deciding. That part stays yours.