Alteryx Thinks AI’s Future Depends on Analysts Not Engineers

Andy MacMillan wants you to stop thinking about AI as a lab experiment and start thinking about it as a business function. As CEO of Alteryx, MacMillan has spent years building a company that helps analysts wrangle data from across an enterprise. Now, in the era of generative AI and agentic workflows, he says those same analysts, often overlooked and under-resourced, are becoming the most critical actors in the AI economy.

“The real missing piece,” MacMillan said in a recent podcast with author and futurist Bernard Marr, “is those kind of business specialists.” Not prompt engineers. Not AI researchers. People who know how their company actually works, how commissions are calculated, how supply chains are tracked, how product metrics are derived and who can systematize that knowledge into AI-ready workflows. “They’re the ones companies are going to turn to when it’s time to actually implement AI.”

It’s a sharp contrast to the dominant narrative, which often centers on model breakthroughs and foundation AI labs. MacMillan’s view is more grounded, but no less transformative: if enterprise AI is going to move from demo to deployment, it will be business analysts empowered with the right tools who make that happen.

And in Alteryx’s world, those tools already exist.

Since its founding, Alteryx has built software that helps non-technical users, financial analysts, operations leads, RevOps managers pull data from different sources, blend it together, and build automated workflows to generate insights or trigger decisions. Originally, this meant liberating analysts from Excel hell. “We’ve always helped business analysts become sort of superheroes and super users of data, or data knowledge workers,” MacMillan said.

Today, the problems are bigger, and the stakes higher. With data sprawled across cloud data warehouses, SaaS applications, and legacy systems, analysts need a workspace or as MacMillan calls it, a “canvas” to orchestrate the mess. But what was once a back-office function is now central to the future of enterprise automation.

As companies move to integrate AI agents into business processes, someone has to feed those systems structured, contextualized data. MacMillan believes that responsibility will fall to a new kind of role: the AI analyst. These are the same business-side operators who’ve always known where the data lives and what it means but now they need to prepare it for AI.

“You might have to grab quota attainment from your CRM, get the commission plan from your commission platform, base salary from your HR system then do the calculation. Someone’s going to be asked: go make the AI do that,” he said.

That someone won’t be an ML engineer. It’ll be a business analyst with tooling from companies like Alteryx.

This marks a deeper shift in how enterprises think about data. Most internal data systems were designed to run the business, not power generative models. ERP and CRM systems are structured around transactional integrity, not semantic queryability. Even data lakes tend to replicate these architectures. But AI requires data to be reorganized around how decisions are made, not how transactions are recorded.

“If you were building an agentic workforce, you might organize the data quite differently,” MacMillan said. “We’re about to see the world’s largest data transformation project as companies rethink what they need this data for.”

Alteryx is positioning itself at the center of that transformation not by building models, but by giving analysts the canvas to translate business logic into workflows that AI systems can ingest and act upon.

MacMillan’s thesis is that the AI analyst will become one of the most valuable roles in the enterprise, precisely because they can marry business understanding with lightweight technical skills. “It’s not just about having super technical individuals,” he said. “It’s about the people who know your business. And we need to provide them with the tooling and capabilities to solve these problems.”

Those problems go far beyond writing prompts. They involve data governance, workflow orchestration, and lifecycle maintenance. MacMillan flagged what he calls “prompting debt”, the proliferation of GPT macros and one-off agent flows that quickly become outdated or unmaintained. Someone has to manage that operational sprawl.

He also emphasized the trust gap that haunts many AI initiatives. “I talk to executive teams with a top-down mandate to use AI and a top-down mandate not to expose any sensitive data,” he said. “And they realize: those two things are incompatible unless you build process.”

Alteryx, once pigeonholed as a data blending platform, now finds itself stepping into a broader role: not just enabling workflows, but helping define enterprise AI strategy. That includes building “clearinghouses”—control points where data usage, access, and business logic are standardized before AI agents can interact with them.

To support this, the company is doubling down on AI-native features across four areas. First, it’s helping customers prepare and transform data specifically for AI applications. Second, it’s building a natural language co-pilot to help users generate workflows without writing code. Third, it’s rolling out “magic reports,” where AI replaces dashboards with narrative summaries of business performance. And fourth, it’s enabling customers to embed large language model calls directly into their workflows summarizing tickets, classifying product reviews, and more.

But none of this, MacMillan insists, changes Alteryx’s fundamental mission. It just raises the stakes. “We’re still focused on the same community,” he said. “We want everyone here to have an AI-forward career path.”

Alteryx users, ACE experts, data-savvy analysts suddenly finds itself in the spotlight. “Their skill set is going to be even more valuable,” MacMillan said. And for the first time in decades, enterprise software may be placing its biggest bets not on engineers or data scientists, but on the people who’ve always understood how businesses actually run.