There's a lot of hype in the marketplace about AIOps right now, and there's a lot of people who've got some interesting ideas about what it should be. The most common idea that I hear is that it's essentially a layer of AI magic that sits across everything that you've got in your IT tooling today and then make sense of all of that for you and then we'll decrease the number of incidents you have and reduce your MTTR...

Imagine this, you are driving your car and you have a tire pressure monitor, works pretty well on three out four tires, you can never quite get it working on the other one. It doesn't always kind of report when your tires are just starting to go down, but always reports when they've just totally gone down.

Then your oil level light, yeah, that's kind of okay, although it blinks on and off now and again, for some random reason and you've got several other monitors in your car. And then what you're trying to do is replace your dashboard with something that says, take all the data from those things that report in different ways and don't always do the job perfectly and they might measure it five minute intervals and one minute interval in different types of data or whatever. Take that and then make sense of it and give me all the answers, you can see what the problem is here, right? You're not feeding that dashboard with great data and therefore you're asking AI to make sense of essentially a fair mess and come out with some solutions. It's going to come up with some good stuff, right? It will say, "Oh yeah, I can see your oil light is on and you've got a flat tire, don't drive your car right now, take it to the gallery to get it fixed," but it's really not the answer to them problem. The problems are underneath.

So if you're thinking about AIOps, you really need to be thinking about what are the outcomes that you're trying to achieve. Firstly, as a business, what do you need to move the needle on? Where are you with that data? And can you fix those data points and then really move the needle on things rather than just putting a mask over the top.