Why AI ticket triage saves seconds, not your MSP

Welcome back to this series all around AI. So far I've gave you a slight intro into AI. We've spoke around some of the technical jargon that you're going to come across and what it kind of means in English. I've probably butchered some of it. And today I really want to speak about what does and doesn't work in my opinion when we're thinking about AI.

Now before we get into some of the things that we've used it for, um, there will be some future videos around actual application. I want to speak about something that has been on my mind ever since AI came out and that is what problem are we actually trying to solve? What are we taking away from people versus what are we giving them? Right? It needs to be a trade, a balance.

I watched a video recently that Rory Sland said about Dyson, um, and how they, you know, all the industry or they were trying to bench the success of their support on how quick someone could come off the phone, and then Dyson, Mr. Dyson, said we're looking at it wrong. We should be grateful our clients are ringing us and looking at it as an opportunity rather than as a problem.

And it's a great segue into kind of what I think about AI really in our space, especially. So let's break down a few fundamentals. What do we know about our space? Well, our clients, or MSP clients to be specific, want three things from you. Really, three things from you. Um, as an MSP, they want to know if they're stuck and really need help, you're there to help them, okay? They want to go to bed at night knowing they're um safe and secure, right, cyber-wise. And they want to know if they need future advancement and help in technology that you're also there with their journey. Right? Three basic functions of what an MSP does for its client.

Then what do our staff at our MSP expect from us, right? As leaders, as owners, as management. Well, our staff want opportunities to grow, okay? They want opportunities to um progress in their career and have development plans. Um, they want opportunities to try new things and to fail, right? Is that what they really need from us as management.

Now, how do we keep both of those things in mind, but also think about the adverse effect that AI could have on those things? Well, let's talk about AI chatbots to start with. So let's talk about a live chat. If we put live chat in front of your clients or in front of our clients and they are in that first phase of I'm really stuck, I have a board meeting, I need to print some documents out, and then they get kicked through either an AI live chat or they log a ticket and AI starts to debug it, they're going to get excruciatingly frustrated with that.

Now, I think the human element of support is the key to its success. Disagree with me all you want. I would always prefer my guys to pick up a phone and ring someone. Yes, it's less efficient, but as Dyson acknowledged years ago, and I've always had this belief, ringing a client and helping them opens more doors than anything else in your business. You not only fix their problem, you build personal connections. People buy from people. It lets you know that they're happy. You discover other problems. There's upsell opportunities. There's so many things that are driven from a quick phone call, okay?

So we flip it on its head and go, right, we're not going to use AI to solve the client's problems, but we're going to use AI to solve our problems. So when a client logs a ticket, AI can go search over all of our documentation and provide us the answer, right, on what to do.

Speaking from experience as a technician, if 90% of the tickets I got already had the solution, I would be bored out of my mind. As someone who has worked through the ranks at an MSP, the most rewarding part of the job is fixing the problem. But to fix the problem, you need to go on the journey of identifying the problem, finding the answer, trying things, and then fixing the problem. If we put AI in discovery and research and searching our knowledge base, then the technician just has a problem to fix. Well, you've really taken away all the things that make them love their job.

You also have to think about it from a psychological impact on what you're doing to the people on your desk, right? They're no longer um having critical thinking skills improved on. They're no longer troubleshooting jobs. They're no longer failing as much because the answer is presented to them. So you go, great, don't worry about it, Connor, I'll get rid of all my level one engineers and I will let AI tell the client what to do to fix it or to run scripts to resolve the problem. If you genuinely think that's going to be something you want to put in your business, I think you are mental.

I don't want in my life an AI automatically responding to my clients. I certainly don't want it ever running scripts against endpoints, even if the only script was a ping for God's sake. Like, I don't want just no. I didn't even trust myself running scripts I wrote, let alone an agentic AI finding a script and running it itself. That to me is absolutely crazy. Like, I just there's no way I'd ever see that.

So what is it that AI can do for us? If we don't want it chatting with our clients and we don't want it providing us all the answers, then what is it we doing? Well, we need to think about it as a tool. We need AI to remove the things that we hate doing or we dislike doing or that takes us a load of time like admin that it can help us do.

Here's a prime example. If you're having phone calls with your clients, you should be recording them all. You should be throwing them through a transcription service and they should be writing your ticket notes. No one likes writing notes, but we like solving the problem. Do that, you do that. That will save you loads of time. It will stop any frustrations of your guys writing the notes and will give you way better output and quality than what your technicians are doing right now.

I don't want to spoil this too much because there's a video on it. But let's take QBRs. Um, Halloween special that's coming by the way. You will see um a full video of our QBR system. But let's say QBRs. One of the biggest problems we had with QBRs in the space is how long it takes to go and gather all the data from all the systems and pull it all together. Well, AI is really good at that, right? You could throw it all of your help desk tickets, all of your projects and where they're currently at, any other systems, pull it all together and give you a summary. Absolutely fantastic. The time saving is great. We still have to go through it as a human, so we can validate it. We could ask it for sources. Where did you find this ticket? Give me the ID. Can save loads of time. What we like doing, as you know, in sales or as QBR is actually meeting with the client, sitting down with a coffee, chatting about their problems, helping them solve solutions. Not many of us love the laborious task of data aggregation.

What about ticket categorisation, Connor? You spoke about it earlier. What are your thoughts on it? I think it's kind of a waste of time if I'm being honest. Now, don't get me wrong. I don't think any of us in this space loves categorising tickets or triaging tickets in. I think AI can do it fairly well, but how much time is it saving? Seconds for the most part. Like, I don't know what you're doing with triage, but is it an incident or request is where we go, give it a base category, give it a priority or impact and urgency. I think that takes less than 15 to 30 seconds around there, right. I don't think the volume of tickets I'm triaging on a day we're offsetting with AI triage are particularly worth it.

However, what is really good about it is that instant triage of impact. So a prime example of this is if we, I'm always, I'm a big believer by the way in humans triaging tickets, not dispatching but just triaging themselves. Um, I think if an AI can take the ticket on inbound and give it a criticality or a priority rating, that is really important. So we are instantly having a desk where the priorities are set. So we are catching P1, P2 tickets quicker. I think that is really valuable.

I think AI triage is a nice party trick to have. I think it is helpful. I think you get probably slightly more accurate categorisation of tickets, but if I was to rate it as massively impactful versus not impactful to a business, I think it's something we can do. It's great. I think if you do or don't have it, you're not massively ahead of your competition and you're not massively falling behind. It is what it is, basically.

Um, sentiment analysis again, right? Let's talk about sentiment analysis. Does that work? Not really. Conceptually, sure, right? We can know if a client is feeling um annoyed through email transaction, right? But the problem is, is who is managing that? Do you have a problem with all of your clients getting pissed off because you speak to them poorly? Isn't that actually a training issue rather than an AI spending money and looking at all of your data problem?

Um, I don't, we've used it a couple of times for companies where there was a problem to raise, really or properly understand, but I don't think it's amazing if I'm being honest with you. Um, summarising tickets is useful. So if a uh ticket has been fully worked to give an internal summary of what was done is not too bad. Um, it's okay. Um, but the point is is that we need to be thinking about AI as a tool. We think about it, of what can it help us do quicker and better, right? Not how do we get rid of our staff?

And the question you should really ask is what parts of your job are you really not enjoying doing? Can we leverage AI to help with some of the inputs? Not take it all away. If you're in a role that you hate all of it, you're in the wrong role. AI isn't going to fix that for you because you'll still hate it. Even if you get AI doing all of it, you'll actually find more frustration because you're lacking the inputs to get this final thing done, right? So you'll, I think you'll hate it more.

So I think you've really got to look at what problem it is you're trying to solve. And that for me is the takeaway of this section, as in what works, what doesn't work, is what problem are you trying to solve? Stop trying to strong-arm AI into your business. Try and look at it as a tool, and if you're saying oh, we have this problem, think could AI help us with that problem rather than we have AI, where do we use it? So that is my section on what works versus what doesn't work in AI for MSPs in our service offering. Thank you very much. Speak to you all soon.