Exclusive: Appian's Marc Wilson on why AI needs process to deliver business value

Techday NZ

16-05-2025

Artificial intelligence doesn't work in a vacuum. That's the key message from Marc Wilson, founder and Chief Executive Ambassador at Appian.
He believes businesses are still struggling to generate value from AI because "they're not integrating it properly into their core operations."
Speaking with TechDay during a recent interview, Wilson offered a blunt assessment of the current AI landscape in enterprise.
"One of the biggest misconceptions that most businesses and most organisations have about AI is that it's indistinguishable from magic - that it just shows up and solves everybody's problems," he said.
According to Wilson, too many companies treat AI as an end goal rather than a tool to improve specific outcomes.
"I've heard time and time again, senior leaders in organisations basically coming to us and saying, 'I have to deploy AI,' as if that's an end state. The truth is, if you don't look at AI through the lens of value, it's indistinguishable from a science experiment."
Appian's core philosophy is clear: AI works best in process. Wilson emphasised that for AI to drive change, it must be "operationalised" - embedded directly into the workflows that govern how an organisation functions. "For an AI capability to affect change in a positive way, it needs to plug into one of those operational flows," he said.
"A good example here in Australia is our work with Netwealth," Wilson said. "They used Appian to orchestrate how client service requests were handled, embedding AI to classify and route customer emails."
"They achieved 98% accuracy - and got the project running within minutes."
Wilson highlighted Hitachi's efforts to unify customer and sales data from across its hundreds of operating companies, and Queensland's National Injury Insurance Scheme, which used Appian's generative AI to extract data from documents with 100% accuracy.
Appian also recently launched its new Agent Studio platform at the event, introducing what Wilson described as "agentic AI". Unlike standalone tools that execute isolated tasks, Appian's approach allows AI agents to function as structured contributors within business processes.
"With our agentic studio, we're able to tie agentic AI into larger, meatier processes - tasking agents the same way you'd task people or systems," Wilson said. "We're combining multiple agents into an overall journey."
That structured approach, Wilson argued, is essential to scale AI safely and effectively. Without a clear framework, he warned, AI agents risk becoming uncontrolled or ineffective. "More organisations are going to get very frustrated very quickly, because they're just going to have this agent, they expect it to do something, and they'll prod it and hope," he said.
"If it's not tied into a structure, there's a lot that can go wrong."
Governance, he added, must be built in from the start.
"Governance and structure are going to become increasingly synonymous," he said. "This is what processes you're allowed to call, what data you're allowed to see, and the limits of your actions. I've created a circle that within it, the AI can do lots of things, but I've constrained the inputs and outputs."
Another critical piece is data. AI's performance depends on access to high-quality, integrated information - but that's a challenge when data is spread across disconnected systems.
"One of the problems that most organisations have today is that a lot of their data is siloed," Wilson said. "Those silos stop really good AI development and learning."
Appian's solution is its patented data fabric, which allows data to be accessed and written across disparate systems without physically moving it.
"It creates a virtualised database, allowing you to consolidate customer data and write back to systems," Wilson said. "The AI capabilities come along with that."
Wilson is clear about the risks of poorly integrated AI. There's the obvious threat of rogue agents making unauthorised decisions, but there's also the quieter failure mode - when organisations fail to realise any return at all.
"If you can't integrate it effectively, if you can't bring it to your processes that matter, it's going to be something that people look at in a year or two and say, 'Yeah, that was a lot of hype, and it really didn't deliver.'"
For companies still waiting to see ROI, Wilson had a simple diagnosis: "That's probably an organisation that's trying to stand up AI by itself, looking at it, waiting for it to produce something without having it truly integrated."
His advice? Start small, and start practical.
"Identify a core business process and think about how AI can remove friction, add speed, or cut costs. We've seen AI take something that took 50 days down to five hours."
And if it feels a little mundane? That might be a good sign.
"Some of the most impactful AI today is going to be boring - and that might be exactly what you want to get started on," Wilson said. "Boring becomes interesting when it drives real value."