The most common subject heading in a technology journalist’s inbox features AI. The stories come in many different guises but at its core, AI sits. When Intelligent CXO sat down with Andy MacMillan, CEO at Alteryx, a leading AI and data analytics company, it provided the opportunity to explore those recurring questions – why don’t people trust AI, is there an AI bubble and can humans and AI really work in harmony or are more redundancies on the horizon?
What is your view on the explosion of AI use within organisations over the past few years?
It’s a fundamentally new capability, so I’m not surprised that people are running towards it and trying to figure out how to use it and how to make it useful and productive. Like a lot of new innovation, I’m not sure most organisations are very good at using it yet, but I think we will be. It is going to be a meaningfully important technology. We’re learning how to wire it into business processes and how to have accountability and governability around it. It’s changing the way we do things here at Alteryx; how we how we write code, how we run our business, how we do analytics. It is a meaningfully new set of capabilities.
Why is there a widening gap between AI ambition and operational reality, and what can be done about this?
Part of it is that we are in this early phase. On one hand, we’re digesting and learning how to use this new capability. We’re changing how certain functions operate. But at the same time, there’s all this discovery and innovation happening on the upside. Every time we feel like we’re closing that gap a little bit, all this new capability comes out, and the gap widens again.
We’re just in the early, high innovation phase of a brand new kind of capability. As we continue to see new capabilities roll out, we’re going to continue to see the gap widen. And then we’ll learn how to use those capabilities. That’s true of a lot of technology innovation. It was true when the Internet came out. It was true when email rolled out.
Why do people not trust AI?
Well, first off, I appreciate that people don’t trust AI. It’s nice that we’re sceptical a little bit when new stuff comes out, and maybe the Internet has taught us to be a little bit more sceptical. Over time, we will, and part of it is learning what it’s good at and what it’s not good at so you start to learn where to use things and where not to use them.
Right now, it is that Jeffrey Moore adoption curve problem, where we have the early innovators saying they can do everything with AI. We have an early majority coming over and learning to use it a little bit. But for a lot of people, it’s still very black box. We don’t know how it does what it does, and it doesn’t give us the right answer all the time. It’s perfectly happy to give us a really confident wrong answer. And most people, when you get a really confident wrong answer once or twice, you stop asking for another answer.
We’ve got to learn as organisations how to how to train it, how to help it give the right answers, how to tell people where the answers came from, teach people how and when to use it. We will get over that trust gap over time.
What are your tips for businesses to ensure AI delivers measurable and scalable business impact?
This happens a lot with technology, where it starts off in the hands of the techies. And I’m one of the techies, so it starts off in the IT-techie universe. The challenge there is, the IT techie universe doesn’t always understand, or usually understand, the nitty gritty depths of the business. For example, how a company manages their tax and audit process is not something most people in the IT department spend a lot of time understanding. For technology to be really valuable is when it crosses over that boundary.
We’re still treating it like a project that IT is going to do. I use the example a lot that in the early days of the Internet, most companies had a website that was entirely run by their IT team. And it didn’t really do commerce. It didn’t really do marketing. It didn’t really do a lot. It was just this presence on the Internet. The real value was when the marketing team was able to use it to actually share information with customers, or the merchandising team was actually able to sell stuff online in a way that was convenient for customers. We’re still in that transition right now.
Do you think there is an AI bubble?
Yes, but I think in the way bubbles often happen in technology, which is at a broad base, there will be a bunch of overvaluation that happens around this. But as always in technology bubbles, there are specific examples of where that valuation was completely justified. Technology goes through this sector interest, and everybody piles money and resources and interest into it, which forms a bit of a bubble. But out of that, there are usually these massive winners that are very successful. There will both be amazingly valuable new companies and industries and things that come out of this AI wave that will be the antithesis to ‘it was a bubble’. There will be a ton of investments that don’t turn into that, or companies that don’t make it over the wall, or things that people got wrong about AI. And there will be real losers in that.
It’s often said that AI will not take people’s jobs away but allow humans to do more meaningful work instead. What are your thoughts on this?
It’s similar to the answer on the bubble, which is, it’s a yes and no. In aggregate, AI will change the constraints on how a business operates. We’ll be able to do things we weren’t able to do before. This happened with the industrial revolution. We changed a lot of the big constraints around how we built things and the manual labour required. We ended up building lots more stuff. The Industrial Revolution created tons of jobs, but it changed the constraints where the jobs were, how companies were organised, and that causes a lot of displacement in the process.
You could say for an individual person, it’s quite likely AI will impact what they do, how they do it, maybe what company they work for, how their company is organised, how many people are in the kind of role that they’re in at that company. Absolutely, AI will impact that. Do I think it causes long-term mass unemployment, because we don’t need people to run companies? I don’t believe that at all. We find a way to always build more companies and more things and more productivity in our society.
I think anybody who feels like AI won’t be disruptive isn’t maybe taking the lens of the individuals that are in the economy versus the economy overall.
And how does your company help other companies with regards to AI use?
We help companies build a lot of automation into what they do. Alteryx is used a lot by supply chain teams, operations teams, finance, accounting, tax; folks like that. I always think of the people in your company that, in reality, sort of run the business on Excel.
We help a lot of people in the business who today solve problems using data, and again, are the spreadsheet crowd, to build the interfaces that work with AI, and they still own those interfaces. The first step of using AI is you have to have predictability and automation in your systems. It goes to your very first question about trust. If I ask the AI to help me forecast my demand planning process, and it has to know how many parts are in the tool shed, I have to believe that it knows how many parts are in the tool shed. If it just guesses, then this isn’t very helpful demand planning. We help people put that infrastructure in place that empowers the business. And then we’re helping people connect their data and their knowledge into these AI systems in a way that the business can drive.
And finally, what’s next for your company?
For us, we’re really focused on helping people automate a lot more of their internal systems and get their data AI ready. And then we’re starting to see people look at ways to agentically build. If you today look at things like Claude code, which is a big trend in AI, where it can write software, the challenge for a lot of people in the business is that’s a really valuable skillset for AI to have. But I don’t speak software development. Our business users don’t want to read Java or Python at the end of this process, but they do want to be able to create things. We’re working with the these big AI vendors to enable AI to instead build them an Alteryx workflow. So instead of ending up with code, they end up with this canvas that explains a workflow of things to interoperate with.
