Straight answer: ChatGPT or Claude can handle most content, research, and admin tasks for £16 to £25 a month, and that's enough for a huge chunk of small businesses. Where it stops working is anything that needs your systems connected together, your data protected, or a decision made about what to automate first. That gap is where a consultant earns their fee, and it's smaller than most agencies want you to believe, but bigger than most DIY guides want you to admit.
Related reading: How Do You Become an AI Marketing Consultant? A Straight Answer From S.
The question I get asked in every single consultation call
Almost word for word: "Can't I just do this myself with ChatGPT?"
Yes. Often. I say this even though it's not in my financial interest to say it, because I'd rather you trust the one piece of advice that matters than doubt everything I say afterwards. A one-person consultancy, a local tradesperson, a solo copywriter, most of them can run their entire AI use on a £20 a month subscription and a bit of patience.
But I also watch business owners spend eight months "getting round to" the AI project that would have taken a consultant three weeks, and that delay costs more than any invoice would have.
What ChatGPT (or Claude, or Gemini) can do for you, on its own
No hedging here, this is the real list of jobs a £20 subscription handles well:
- Drafting first versions of emails, proposals, job descriptions, social posts
- Summarising long documents, contracts, or customer feedback
- Turning a voice note or messy bullet list into a structured blog post or LinkedIn update
- Basic customer service scripts and FAQ answers
- Market research, competitor scanning, and generating angles for content
- Building a simple pricing calculator's logic or working out formulas for a spreadsheet
If you're building something interactive for your website, like the interactive calculators I've written about before, ChatGPT can write the logic and even the code for a basic version. You don't need a developer for a first draft, you need one to make it look trustworthy and load fast.
Where DIY quietly falls apart
Here's the story I tell in workshops now because it's a real one.
A client of mine, a five-person accountancy firm in Surrey, decided in early 2025 to "sort out AI" over a weekend. The owner set up a ChatGPT account, wrote some prompts for client emails, and felt brilliant for about three weeks. Then reality: their client data was going into a chat window with no audit trail, two staff members were using different tools with different prompts producing different tones of voice, and nobody had checked what the firm's insurer or the ICO would say about feeding client financial details into a general AI tool.
Nothing had gone wrong yet. That's the trap. DIY AI usually looks fine for months, right up until a client asks "where does our data go" and nobody in the business can answer. That firm ended up paying for four hours of my time to build one document: a simple usage policy, an approved tools list, and a template prompt library. Four hours. Not a six-month retainer. But it needed someone who does this daily to spot the gap, because the owner, understandably, was too close to their own business to see it.
This is the uncomfortable bit most "just use ChatGPT" articles skip: the tool isn't the risk. The absence of a decision about the tool is the risk.
What an AI consultant costs in 2026
Real ranges, not vague ones:
- Freelance AI consultant, hourly: £75 to £180/hour in the UK, higher in London, lower outside the M25
- Day rate: £600 to £1,400
- A single project (say, a customer service AI setup or a content workflow build): £1,200 to £4,500 depending on how many tools it touches
- Monthly retainer for ongoing implementation and training: £800 to £3,000
- Big agencies pitching "AI transformation": £10,000+, usually aimed at businesses with 50+ staff, not you
I've written a full breakdown with more scenarios on how much an AI consultant costs, but the short version is this: most small businesses need a handful of focused hours, not a retainer. Anyone pushing a 12-month contract for a business with four staff is selling you their cash flow problem, not solving yours.
The real cost isn't the invoice
This is the part that gets left out of nearly every comparison I've read. The consultant's fee is the visible cost. The invisible one is what it costs you in time not spent on your actual work while you fumble through YouTube tutorials trying to connect three tools that don't talk to each other.
Do the maths honestly. If you bill £60 an hour for your own work and it takes you 15 hours spread over two months to get a basic AI workflow running badly, that's £900 of your own time for a worse result than a consultant could deliver in four paid hours at £120 each, £480 total. You didn't save money doing it yourself. You spent more of it, just less visibly, which is exactly why nobody counts it.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Alex Hormozi makes a version of this point in his own material, that the businesses who scale fastest are the ones who pay to remove their own bottleneck early rather than proving they can grind through it alone. I go through more of what's useful and what's overhyped in his approach in my piece on business lessons from Alex Hormozi, but this particular point holds up regardless of whether you like his style.
A simple test to decide which camp you're in
Ask yourself these four questions honestly:
- Does this AI task touch customer or financial data that would matter if it leaked? If yes, get proper advice before you start, not after.
- Will more than one person in your business be using the tool? If yes, you need a shared system, not individual habits, which means someone needs to design it.
- Have you tried to do this yourself for more than two weeks with no real progress? That's your answer. Stop grinding, get help.
- Is the task a single, contained thing, like writing better email subject lines or summarising meeting notes? Do it yourself, today, for free.
Companies that got big doing things scrappily themselves, like Airbnb in its early days famously photographing listings themselves to fix bad conversion, are proof that DIY has a real ceiling that's worth pushing against. I go into how that played out for them in my piece on Airbnb's marketing strategy. But notice they hired specialists the moment the scrappy approach stopped scaling, not before. That timing is the whole skill.
What I'd do in your position
Start with the free or cheap tool for two weeks. Write down every task you attempt and note honestly which ones you finished, which ones you abandoned, and which ones made you nervous about privacy or accuracy. That list becomes your brief, whether you hand it to yourself or to a consultant.
If the list is mostly "wrote some content, saved some time," you're done, keep going solo. If the list includes "not sure if this is safe," "tried three times, gave up," or "need this to work the same way for my whole team," that's the moment to bring someone in for a defined, short piece of work rather than an open-ended retainer. I lay out exactly how that first conversation should go, including the questions to ask before you sign anything, on my page about working with an AI consultant as a small business.
Ahrefs, a company that now runs huge parts of its content and product operations through automation, still started with founders doing the unglamorous manual work themselves before bringing in specialists to build what they'd already proven worked by hand. There's more detail on how that unfolded in my breakdown of the Ahrefs marketing strategy. The lesson repeats everywhere I look: prove it manually first, then pay someone to make it repeatable.
Napoleon Hill wrote about this a century ago in different language, the idea that no one succeeds alone and that recognising when you need someone else's specific knowledge is a strength, not an admission of failure. It's one of the points I still find most useful from his work, which I cover in more depth in business lessons from Napoleon Hill. The principle hasn't aged even if the examples have.
Free resource: The Retainer-to-Project Pricing Decision Mini-Guide.
Frequently asked questions
Is it cheaper to use ChatGPT than hire an AI consultant?
For single, contained tasks like writing emails or summarising documents, yes, ChatGPT at £20 a month is far cheaper. For anything involving multiple staff, customer data, or connecting several tools together, DIY usually ends up costing more in wasted hours than a few hours of proper advice would have cost upfront.
How much does an AI consultant cost for a small business in 2026?
Expect £75 to £180 an hour, or £600 to £1,400 for a day rate, with most focused small business projects landing between £1,200 and £4,500 total. Long retainers of £10,000 or more are usually pitched at larger companies and are rarely necessary for a business under 20 staff.
Can I just teach myself AI without any paid help at all?
Yes for basic content, research, and admin uses. It gets harder once you're handling client data, want consistent results across a team, or you've spent more than two weeks stuck on the same problem without progress, which is usually the signal that a few paid hours would save you more time than they cost.
What's the biggest mistake small businesses make with AI right now?
Treating the tool choice as the decision, when the real decision is what data goes into it, who uses it, and whether there's a written policy anyone can point to later. Most AI problems in small businesses trace back to that missing decision, not the software itself.
Related reading: Using AI to Chase Unpaid Invoices Without Sounding Like a Debt Collector and AI Tool Costs Explained for Business Owners in 2026.
