A business owner asked me this recently, only half joking: “Andy, with AI doing everything now, do I even need an IT company anymore?” It is a fair question, and I would rather answer it honestly than pretend AI changes nothing. So here is the straight version, including the bits that do not flatter my own business.
The short answer: if you are a sole trader you can probably get a long way on AI alone. For a business of small businesses up to 100 staff, especially a regulated one like a law firm or accountancy practice, you need both: AI plus an accountable IT partner, not AI instead of one. Here is the honest reasoning, including where AI genuinely wins.
Key takeaways
- AI is genuinely useful and we use it every day. It is excellent at drafting, summarising and walking you through first-line fixes.
- What AI cannot do is be accountable, restore a failed backup, secure your network properly, or answer the phone when your email is down at 4pm.
- The real danger of “just use AI” is your company data quietly leaving the business, and confident answers that are wrong, with no audit trail.
- For a regulated firm of small businesses up to 100 staff, the honest answer is AI plus a good IT partner, not AI instead of one.
- We use AI to do more for you for less, not to disappear.
First, the honest bit: what is AI genuinely good at?
I am not going to fear-sell AI to protect my own patch. AI is the most useful tool to land in business technology in years. It will draft an email or a document in seconds, summarise a long thread, explain a confusing error message in plain English, suggest the formula you need in Excel, and talk you through the obvious first steps when something is not working. We use Microsoft Copilot inside Initial IT every single day, and it makes us faster. If your question is “can AI help me get more done,” the answer is a clear yes.
What can’t AI do for your business?
Here is where the “do I still need an IT company” question gets interesting, because there is a long list of things AI simply cannot do.
It cannot be accountable. When AI gives you a confident answer that turns out to be wrong, there is nobody to ring, nobody who carries the risk, and nobody whose job it is to put it right. A chatbot does not own the outcome. We do.
It cannot touch the physical world. When a firewall dies, a switch fails, the broadband drops or a laptop will not boot, AI can describe the problem beautifully and change nothing. Someone still has to swap the hardware, re-cable the rack and get your people working again.
It cannot actually restore your data. AI can tell you what a backup is. It cannot tell you whether yours would genuinely restore under pressure, because a real backup is a tested process, not a paragraph. The day you need it is the worst possible day to find out the answer to that.
It cannot make the judgement call that has consequences. At 4:55pm, when an email lands that might be a supplier and might be a fraud attempt mid-incident, you do not want a generic answer. You want someone who knows your business and will make the right call.
It cannot secure your environment properly. AI can describe conditional access, multi-factor authentication and least-privilege permissions in Microsoft 365. It will not configure them for your business, keep them aligned as you grow, or stand behind the decisions. Security is something you set up correctly and own, not something you ask a chatbot about once. That is the heart of our cyber security work.
And it cannot answer the phone. There is real value in a person who already knows your setup picking up when something goes wrong, instead of you starting from scratch with a search box. That is what managed IT support is really for.
What is the hidden risk of letting AI fix your IT?
This is the part that worries me most, because it is happening quietly in lots of businesses already.
When your team uses public AI tools to “sort” a problem, two risks turn up. The first is data. The moment someone pastes a client matter, a spreadsheet or an email chain into a free public tool, that information has left your business, and you have no idea where it is stored, how long it is kept, or whether it is training someone else’s model.
For a law firm or an accountancy practice, that is a confidentiality and compliance problem, and “we were only trying to be efficient” is not a defence the SRA or the ICO will accept.
The second risk is the confident wrong answer. AI does not say “I am not sure.” It gives you a fluent, plausible answer that is sometimes simply incorrect, and if you act on it with your live systems, you can turn a small problem into a big one, with no audit trail of who changed what or why.
None of that means do not use AI. It means use it inside guardrails, which is exactly the work a good IT company does now. The National Cyber Security Centre says the same thing in its guidance on using AI safely at work.
How do AI and a good IT company work together?
We are not anti-AI. The opposite. The most useful thing we do for clients this year is help them use AI safely, so the whole team gets the benefit without the data leaking. We roll out Microsoft Copilot inside your own Microsoft 365, where your information stays in your tenancy and never goes off to a public model. We put a simple AI policy and the right controls in place so people can be productive without putting the business at risk.
AI also makes our own service better. We use an AI service desk built for IT companies like ours to triage incoming support tickets and help write up our knowledge base articles. In plain terms, that means your request gets read, sorted and routed to the right engineer faster, and once we have solved something the fix is documented and reused instead of being worked out from scratch next time. The AI handles the repetitive sorting, and our engineers spend the saved time on the work that actually needs a human.
So you are not choosing between AI and an IT company. With the right partner you get both: the speed of AI and a person who is accountable when it counts.
So, do you still need an IT company? The honest answer
It depends on you, and I will give you the honest version rather than the self-serving one.
If you are a sole trader with a laptop, a phone and a couple of cloud apps, you can probably get a long way with AI and a bit of common sense, and you may not need a monthly IT contract at all. That is fine, and I will tell you so if you ask.
If you are a business of small businesses up to 100 staff, especially a regulated one like a law firm or an accountancy practice, then the answer is yes, and arguably more than before. Not because AI is not useful, but because the risks AI introduces, around data, security and compliance, now need governing by someone who is accountable. The businesses that win with AI are the ones who put guardrails around it, not the ones who let it loose and hope.
That is the conversation we are happy to have, even if the outcome is that you do not need us yet. We work with growing firms across Lichfield, Staffordshire and the West Midlands, so if you are nearby and not sure where you stand, book a quick chat, or have a look at what our support actually costs.
Free 2-minute tool from Initial IT
Should you roll Microsoft Copilot out yet?
11 honest questions, an instant readiness score, and a short plan tailored to your business. No marketing list.
Take the readiness quiz →Frequently asked questions
Can’t I just use ChatGPT or Copilot instead of an IT support contract?
For day-to-day “how do I” questions, AI is a great help and you should use it. But it cannot fix hardware, secure your network, restore a backup or be accountable when something goes wrong, so it complements an IT partner rather than replacing one.
Will AI replace IT support and IT companies?
It will change the work, not remove the need. AI is taking the repetitive first-line questions, which frees good IT people up for the security, planning and judgement work that actually protects a business. The accountability still has to sit with a human.
Is it safe to use AI tools in my business?
Yes, if you use the right tools in the right way. The danger is public free tools where your data leaves the business. We set clients up with Microsoft Copilot inside their own Microsoft 365 so the benefit stays and the data does not leak.
We are a small business, do we really need IT support?
Maybe not, and I would rather tell you the truth. A one or two person business can often manage with AI and good habits. Once you have a team, client data and compliance to worry about, the maths usually changes.
How does Initial IT use AI?
We use Microsoft Copilot internally to work faster, and an AI-powered service desk to triage tickets and help build our knowledge base, so your request reaches the right engineer quickly and common fixes are written up once and reused. We also help clients adopt AI safely. The point is to give you the upside of AI without handing your data to a public model or losing the human who picks up the phone.
– Andy Price, Founder & MD, Initial IT · 01543 524594 · hello@initialit.co.uk