“One of the major problems with AI today is that enthusiasm is getting ahead of practicality, and people are doing AI projects for AI’s sake.”
This observation comes from Marc Wilson, Appian Founder and Chief Executive Ambassador. Dynamic Business sits down with Wilson to discuss the risk of organisations investing in AI, automation and low-code platforms but overlooking one critical element: process.
Business modernisation is built on structured processes
“When you take a step back and think about what you’re wishing for, most organisations are wishing for more efficiency, more cost savings, a better constituent or customer experience,” says Wilson. “And then you ask the question, what process is required, and how can I put AI into that process to make it better?”
Wilson argues that AI investments often stall before delivering ROI because organisations aren’t first considering how they are going to deploy the technology.
“Thematically, one of the things that Appian is driving in the market is the idea that AI needs process, and I would extend that a little bit further to say, AI needs a process in order to show value,” he explains.
“The reasons for that are pretty clear: you can create an AI algorithm or agent that magically does X, Y, Z. But if there isn’t a way to plug X, Y, Z into how the business operates, it’s not amounting to much.”
Seventy per cent of Appian’s customer base currently uses AI, and Wilson says that in a significant number of cases, it is not being applied in a new greenfield project but rather integrated with processes that have already been automated.
“One of the most widespread useful examples of AI that we see is in email extraction, classification and routing. We work with a lot of organisations that still get a substantive amount of their customer or vendor data coming in via email. And they have people assigned to read those emails and to send them in the right direction, or maybe they have older tools that are designed to figure it out.
“And there are high failure rates in that, because emails come in all different shapes and sizes. Applying AI skill sets that aren’t focused on rule sets but on a generative approach to understanding the context of the email, the context of the document is proving to be enormously valuable. But let’s face it, that’s the boring side of AI. Who wants to talk about email classification? But to most businesses, that’s value.”
Read also: How data fabric can turbo-charge your business transformation
AI makes businesses smarter and more responsive
As a technology partner, Appian’s focus is on helping customers deploy AI and automation within structured processes to ensure they doesn’t become disconnected tools with limited impact.
One such company is fast-growing Australian wealth management company Netwealth which is using the Appian Platform and AI to automate its email ingestion process for client and advisory case management systems.
Queensland’s National Injury Insurance Scheme (NIISQ) is another Appian client using AI for data extraction from incoming documents. Wilson says that speeding up NIISQ’s processes is particularly important because participants have experienced catastrophic accidents and are enduring some of the worst days of their lives. Instead of people having to go through documents and laboriously work out particular data points, introducing an agentic capability to data extraction has sped up the application process and achieved virtually 100 per cent accuracy.
AI facilitates a richer human experience
Embedding AI into process workflows results in better case management, faster compliance and more responsive services without compromising data privacy or workforce confidence.
Wilson says that many organisations, especially highly regulated institutions like banks and insurance companies, are integrating AI into their customer journey to deliver a richer human experience.
In an academic setting, The University of South Florida has pioneered the use of generative AI to improve its advisor community’s ability to provide support to the student body through process orchestration.
“Process orchestration, for us, is a situation where many different activities are going on around a particular topic, in this case a student,” explains Wilson. “Process orchestration helps bind and tie the student journey together, and advisory services are a part of that. Here, the AI components are about bringing a rich set of tools to allow the AI to enhance the human experience.
“Not only is this giving the advisor a consolidation of the data, but it also provides an opportunity for advice on what can be discussed, maybe calling out particular areas of concern or optimism.
“What AI does is make the prep time easier and more valuable going into the advisory session. The advisor can be better, more personable, and more engaged. They can be more involved because they’re better prepared for what they do.”
Building trust in AI through human oversight and governance
Despite its exciting capabilities, concerns about AI technology remain. “Most organisations are not willing to have an AI run its operation,” says Wilson. “The people, the management team will run the organisation; they’re looking for AI to help them out.
“Process creates a safety net. Horror stories are emerging, and we’re going to see more examples of AI running amok. They’re doing things they shouldn’t: erasing data here, making an order there, doing something the company doesn’t want done over here. This occurs because the AI is unbounded: it wasn’t told what it could and couldn’t do, but a process creates a safety layer around it.
“AI within a process also gives organisations the opportunity to create an audit trail of what was asked of an AI and what an AI did. Because you can tell an AI to make a decision but then use this process to do what you want, everything becomes a line item in an audit trail that becomes easier to diagnose and figure out and see what was actually done. And, of course, it gives you a much better way of scaling a lot of these distinctive AIs because there’s more confidence when there’s safety. All these things are about giving organisations confidence to understand what AI can do, what it should do, and perhaps most importantly, right now, what it shouldn’t be doing.
How Private AI protects data
Wilson weighs into the Public AI vs Private AI debate by highlighting some important benefits to the Private AI option: one that uses a model as a starting point but is then trained by an organisation’s data and information.
“First, I think the privacy concern is paramount. If the large model that everybody’s using is being fed your customer information, your constituent information, your patient’s health record, there’s a leap of faith that I think not too many organisations want to take.
“I have yet to meet an organisation in my travels around the world, that is comfortable with pushing its data into a public cloud environment. And a lot of that starts with the privacy concerns. It starts with concerns about the rules and regulations that might exist in a particular state or at a national level.”
Wilson says that companies have legitimate misgivings about sharing data with competitors. “Most banks don’t want to train a model to help their competitor out. They want to train the best model for themselves. Governments don’t want to be in a position where their constituents’ personal information is being mixed with another country’s constituent personal information.”
He believes that most AI infrastructure is moving to a private model. “The best analogy I can give is you go to a store, pick your toy off the shelf, and pick your AI that’s been trained to do something off the shelf. You take it home, and you begin to add your own information to it. And you make it better. But you are making it better based on your own information, your own likes and dislikes.”
Keep up to date with our stories on LinkedIn, Twitter, Facebook and Instagram.
