How entrepreneurs can use AI to run a leaner business

Starting a business used to mean a hard choice: do everything yourself, often badly, or hire help you could not yet afford. AI changes that sum. Used well, it gives a founder something close to a team of willing assistants: a researcher, a copywriter, an analyst and a coder who never clock off. Used carelessly, it produces confident nonsense at scale. Here is how to get the first without the second.

What AI is good at, and what it is not

The AI assistants most people use today are built on large language models: systems trained on enormous amounts of text to predict what a useful answer looks like. That makes them remarkably good at some jobs and surprisingly poor at others.

They are strong at drafting, summarising and reshaping text, at generating lots of ideas quickly, at explaining unfamiliar subjects in plain language, at working through structured information such as spreadsheets, and at writing computer code.

They are weak at facts they have not been given. A model can state something false with complete confidence, a habit known as hallucination, and it will happily invent a statistic, a quote or a source that looks entirely real. It also knows nothing about your business, your customers or your standards unless you tell it, and it has no judgement or taste of its own.

The most useful way to think about it is as a very well-read new starter on their first day. Enormously capable, eager to help, and in need of clear instructions and someone to check the work.

Research and thinking

A surprising amount of a founder's week goes on reading. AI can take a lot of that weight.

Paste in a long report and ask for the five points that matter to your business. Give it the notes from ten customer conversations and ask which problems came up most often, in the customers' own words. Hand it your competitors' product pages and ask how each positions itself, who it seems to be for and what none of them is saying.

It is also an excellent sparring partner. Describe a plan and ask it to argue, as convincingly as it can, why the plan will fail. You will not agree with every point, but you will usually find one you had not considered, and it is far cheaper to discover that in a chat window than in the market.

Writing and marketing

This is where most small businesses start, and for good reason. AI can produce a first draft of a product description, a welcome email, a landing page or a set of ad headlines in seconds. More usefully, it can produce twenty variations, which is exactly what you need for testing adverts properly.

It is also good at repurposing. A single detailed article can become a newsletter, a run of social posts, a short video script and a set of frequently asked questions, each reshaped for where it will appear.

Two cautions. First, keep your own voice. Give the tool examples of your best writing and ask it to match the tone, then edit the result as if a junior colleague had written it. Second, beware of sameness. If you and your competitors ask the same tool the same question, you will all get much the same answer. Your advantage is the specific knowledge only you have: your customers' real words, your results, your opinions. Put that into the brief and the output stops sounding like everyone else's. Our guide on how to brief an AI tool properly shows how.

Customer service

AI can draft replies to customer emails using your own policies and frequently asked questions, summarise a long complaint thread before you pick it up, and check that a reply sounds as calm and helpful as you meant it to.

It is also a quick way to build a help centre. Give it a month of customer questions and ask it to group them into themes, then draft an answer for each one that you can check and publish. Fewer repeat questions means fewer emails.

If you put an AI chat assistant on your website, keep it on a short lead. It should answer only from information you have approved, say clearly when it does not know, and make it easy to reach a real person. One confidently wrong answer about a refund or a delivery date can cost more goodwill than a hundred right ones earn.

Numbers and data

Few founders love spreadsheets, and AI makes them far less painful. Describe what you want in plain English and it will write the formula, explain how it works and fix it when it breaks. It can tidy a messy export from your shop or ad account, suggest how to chart it, and point out trends worth a closer look.

One rule matters here: let the spreadsheet do the arithmetic. Language models can make mistakes with numbers, so ask the AI to write the formula rather than to calculate the answer itself, and sense-check any figure before you act on it.

Building things without being a developer

This is one of the biggest changes of the last few years. AI coding assistants now let people with no programming background build simple tools: a calculator for a website, a script that renames hundreds of files, a page that pulls figures from one system into another. Experienced developers use the same tools to work several times faster.

Start small, test everything, and keep a copy of anything that works before you change it. Most importantly, understand roughly what you are putting live. A tool that saves you an hour a day is wonderful; one that quietly sends customer emails to the wrong address is not.

Automating the admin

The dull jobs are often where AI saves the most time. Meeting tools can now transcribe calls and produce a summary with actions. Email assistants can sort an inbox and draft replies to routine messages. And automation tools such as Zapier, Make and n8n can connect the apps you already use, with AI steps in the middle.

A simple example: when a new enquiry arrives through your website, an automation can summarise it, add the details to a spreadsheet, label it by type and draft a reply for you to approve. None of that is clever on its own, but together it turns ten minutes of admin into one click.

Choosing your tools

There are three broad kinds of AI tool:

• General assistants such as ChatGPT, Claude, Gemini and Microsoft Copilot, which handle a wide range of tasks through conversation.
• AI built into tools you already use, from Google Workspace and Microsoft 365 to Canva and Notion, which saves switching between apps.
• Specialist tools for particular jobs, such as transcription, image creation, video editing or coding.

We are heavy users of AI ourselves. Claude and ChatGPT are part of our working day, alongside many other tools, and the biggest lesson we have learned is this: the landscape changes so quickly that nobody can keep up with all of it, and trying to is a trap. New models, apps and features arrive every week, each promising to change everything. Choose carefully which tools, models and projects you put your time into, learn those properly, and let most of the noise pass you by. The skill that lasts is knowing how to brief and check an assistant well, not which button lives where. If you would rather keep your data entirely in-house, it is also possible to run AI models on your own computer; our guide to the hardware you need to run AI yourself explains what that takes.

AI will not give you a shorter week

There is a popular idea that AI means working less. In our experience, it does not. What it does is shrink the time each task takes, and the time it saves never sits empty for long. It fills with the next job, and then the one after that. A product description that used to take an hour takes ten minutes, so you write fifty instead of five. An analysis that took a day takes an hour, so you run it every week instead of every quarter.

That is the real prize: not a shorter week, but a business that gets far more done in the same week. It also means you need to decide what the saved time is for. If you do not choose, it will be swallowed by whatever shouts loudest. Point it at the work that moves the business forward: conversations with customers, the next market, a better product.

Where to be careful

• Accuracy. Check facts, figures, names, quotes and links before anything goes public. If a claim matters, verify it from a source you trust.
• Confidentiality. Before pasting in customer details, contracts or anything commercially sensitive, check what the tool does with what you give it. Many offer settings or business plans that keep your data out of model training.
• Over-reliance. AI is a brilliant first draft and a poor final word. Keep thinking for yourself, especially about anything that involves judgement, money or people.

How to start: one task, two weeks

The fastest way to learn is to pick one real job and stick with it:

1. Choose the task you most dread or that eats the most time each week.
2. Use an AI assistant for that task every working day for two weeks.
3. Note how long it takes with and without AI, and how good the result is.
4. Save the instructions that work best, so you can reuse them.
5. At the end, decide whether to keep going, adjust, or move on to the next task.

Two weeks is long enough to get past the novelty and short enough to stay focused. Most people find at least one task they would never go back to doing by hand.

Where to go from here

AI will not run your business for you, but it can take a remarkable amount of the work off your desk. The pattern we see is consistent: the founders who benefit most are the ones who know their customers best, because they give the tools the best material to work with. Learn to brief it well, check everything that matters, and spend the time you save on the parts of the business only you can do. If you are just starting out, our list of the books every first-time founder should read is a good companion.