/ Home / Blog / AI on a small budget
You do not need an AI department. You need one annoying task gone.
Most of what is written about AI is aimed at companies with a budget and a team to match. Small businesses read it and quietly assume the whole thing is not for them. It is, but the way in is different. You do not start with a strategy. You start with the one task that eats your week, and you remove it.
August 2026 ยท Marlow
What can AI realistically do on a small budget?
Quite a lot, as long as you keep the scope small and specific. Answering the same customer questions on WhatsApp, drafting replies and posts, reading documents and pulling out the details, sorting enquiries, catching the order that would have been missed, all of this is within reach without a large project. The trick is that each of these is one job, not a platform. Small budgets do well with narrow, well-defined tasks and badly with vague ambitions to add AI to everything.
Where should a small business actually start?
Start with the task you personally dread, the one that is repetitive, happens often, and does not need judgement. For most owners here that is the same handful of things: replying to routine messages after hours, retyping the same information between a chat, a spreadsheet, and an invoice, or chasing the same reminders. Pick one. If it wastes real hours every week and follows a predictable pattern, it is a good first candidate. If it needs taste, trust, or a difficult human conversation, leave it to a person for now.
What should you ignore for now?
Ignore anything sold as a complete transformation of your business. Ignore tools you would have to bend your whole workflow around. And ignore the pressure to have an opinion on every new model in the news, because none of that changes what your customers need from you this week. On a small budget, breadth is the enemy. One task done properly will teach you more, and earn more, than ten features you never quite switch on.
How do you know it actually worked?
Decide the measure before you start, and keep it simple. Hours saved a week, messages answered outside working hours, mistakes caught, enquiries that turned into replies instead of silence. Write down roughly where you are now, run the change for a few weeks, and compare. If you cannot see a difference in something that matters, stop and try a different task. This is the discipline that keeps a small budget honest: you are buying a result, not a subscription.
What does it cost to try?
Less than people fear, and the real cost is usually your attention during setup rather than the tools themselves. The honest variables are how many systems the task has to touch and how much cleanup your existing information needs, because AI is only as good as what you point it at. A single, well-chosen task can often be set up quickly and cheaply, proven, and only then expanded. That is the whole strategy for a small business: start with one thing, make it pay, and let each win fund the next.
Questions owners ask us
Is AI really worth it for a business my size?
Yes, if you aim it at a specific repetitive task rather than at your business in general. Small businesses often see the clearest return, because a single owner wears many hats and even one task removed frees up real time. The mistake is trying to do everything at once instead of proving one thing first.
Do I need to replace my staff to benefit?
No. The best early uses take away the dull, repetitive work nobody wanted, so your people spend their time on customers and judgement calls. It works best as something that handles the first pass and hands anything unusual to a person, not as a replacement for the person.
Will it need constant babysitting?
A well set up task should mostly run on its own, with a person reviewing the edge cases rather than every action. Poorly set up ones do need babysitting, which is usually a sign the scope was too broad or the underlying information was messy. Start narrow and it stays low maintenance.
Tell us the one task that wastes your week. We will tell you if AI can take it.
No strategy deck, no big project. Describe the repetitive job you would most like gone, and we will tell you honestly whether it is a good first candidate, what it would involve, and how we would prove it worked before you spend more.
Start a conversation