It seems that AI is mentioned in every second sentence right now. AI this, AI that, and AI the other thing.

And we know it gets a bit exhausting. But if you step back from the hype, there are some pretty impressive gains to be made apart from writing emails and LinkedIn posts.

Like what?

The (really) good stuff

One project we’ve been working on for a customer is AI-powered voicemail transcription. To some, that may sound ho-hum, but to us it’s actually pretty interesting.

Say you’re a health organisation, and you get masses of voicemails left on your phone system as you can’t handle them all in real time. So your receptionist, nurse, or whoever gets to listen to every call, grabs the patient’s details from it, notes what they are calling about, then prioritises and acts on it. It may just be a request for a repeat prescription. Or it may be asking for an urgent appointment or blood test results. Or an overdue invoice. Whatever, we all know that feeling of dread when faced with 80 voicemails to listen to and process asap.

These days, lots of these patient requests can be done through an online portal. But the reality is that some people still prefer to pick up the phone, especially older patients or customers who find self-service systems frustrating or too impersonal.

Now this is where AI starts to earn its keep.

Instead of staff having to listen to every voicemail and decide what to do with each one, AI can transcribe the message and extract the useful information. That information can then be emailed to one of the doctors, or to the receptionist to make an appointment – and urgent calls are triaged so they go to the top of the list.

If you’re not a health organisation but handle a large volume of support calls – as Colton does, for instance – you can use it to add a service ticket. So you can imagine how well this would work for insurance companies with a high volume of claims enquiries made by phone, a retailer with a big returns department, or an internet provider.

For a busy team at a health organisation, it may only save five or ten minutes at a time, but across a day, a week or a month, that really adds up.

The bad stuff

One area we’re investing in is AI training and workshops for our customers – and with good reason.

As humans, we’re naturally curious. And we’ve largely overcome our fear of using AI and started trying out all sorts of tools. The problem is that the (usually free) tools and platforms many employees are quietly playing with in the shadows aren’t sanctioned by the business.

And this ‘shadow AI’ is a big problem.

When you use ‘free’ AI, you have no control over the privacy of the information you share. You may have asked ‘how do I get a raise – everyone else on this spreadsheet is paid more than me!’ or ‘predict our gross profits based on this sales spreadsheet’ only to find that the salary and sales data you’ve shared is now and forever in the public domain. Forever.

Shadow AI is a significant security risk. By putting business data, customer information, intellectual property or internal documents into a platform without proper controls, you may be creating a risk you cannot easily unwind. It’s only by using a paid service like Microsoft’s Copilot (with all its inbuilt security controls) that you can trust that your data stays where it belongs.

Your business policy needs to state that the only AI used in the business is the one you officially provide – and enforce it.

The awesome stuff

There are some common misconceptions about AI. Some people underestimate it and only see it as a research tool. Once they learn that it can summarise emails, draft responses, help prioritise work, prepare meeting notes or consolidate information, they start to see the everyday value.

Others go the other way and expect AI to take over their entire workflows without much technical input. Yes, AI can do a lot, but the further you go into automation, integration and custom agents, the more technical knowledge you need.

That’s where an IT partner (like us) can help bridge the gap between a clever idea and a safe, working solution.

One of the best client examples we’ve seen recently was a business that used AI to consolidate data from multiple Excel reports into a single template. What stood out was not just that AI could help, but that the user gave clear instructions, provided the end template, and explained how they wanted the information structured. With a little polishing, they got the result they needed quickly – think minutes instead of the hours it used to take to do the same job badly!

What can go wrong?

Let’s say your policy is: “Customer data must only be accessed by authorised staff.”

Your data governance defines and approves that rule. Your policy awareness makes sure everyone understands and follows that rule. And compliance ensures it meets our privacy laws (and you can prove it if audited).

Where can it go wrong? Say your staff awareness is weak, and when the pressure is on, a temporary intern is asked to dive into and copy your payroll data and share it in an ‘all-company’ email. In that case, you’ve clearly failed the compliance test – even though in theory you have a strict policy in place.

The upshot is that governance without awareness doesn’t work (it’s just a dusty, old, forgotten document full of rules that no one notices or follows). And compliance without governance isn’t sustainable (without policies, you don’t have a consistent way to live up to your requirements – you’re just winging it!)

Together, they’re what turns data governance from a document into something that actually protects your business.

The golden rules!

  • AI works better when you are specific. Tell it what you want, give it context, and set boundaries.
  • Always ask it to double-check its work.
  • Ask for references – but tell it what sources it can and can’t. So, if you don’t want it pulling from random blogs or forums, say so. If you need published, current, or recent documentation and valid working links, specify that.
  • Always check the output. Sadly, AI can and does make things up – it can sound super confident and still be completely wrong. If you’re using it for research, customer communications, policies or public-facing content, you are responsible for what it produces. So – treat it like a junior team member: useful, fast and capable, but still needing review, guidance, and correction.
  • Finally, don’t let AI replace skill development. While it’s a fantastic assistant, it shouldn’t become a crutch, especially for newer staff. People still need to understand their job, build judgement, and know how to work without AI if the tool is unavailable, too expensive or not appropriate.

The last word

Nothing replaces the human touch or human judgement based on experience or knowledge. But…the reality is that AI is moving fast, and the tools are getting more powerful. Used well, they can save you hours, improve service, reduce admin and help people focus on higher-value work.

On the other hand, when used poorly (and without training), AI can introduce security, accuracy, and operational risks.

So our advice is simple: start with the problem you’re trying to solve, not the platform. Look for practical wins. Put security controls in place. Train your people. Check the outputs.

And when you’re ready to go beyond simple use cases, give us a call.