While many companies are going all in on AI, Unisys is taking a more measured approach.
Mike Thomson, who was recently elevated to the CEO role after nearly a decade with the company, said that his personal viewpoint comes down to a few words: ‘don't be a hammer looking for a nail’.
“You don't just bring AI in and say AI is going to solve your problem, right? You have to understand what the business problem is,” he said.
“You have to understand what the return [on investment] is to use certain emerging technologies to actually fix the problem.”
Thomson’s view echoes that of Brett Barton, the company’s global AI practice leader, who told techpartner.news last year that AI is the first technology that can cause harm to an organisation if they're not careful.
“Clients have blocked ChatGPT, only to have employees take information on their phone and run it through just to see what comes back. And that information is now out in the wild. It's gone forever," said Barton.
“And for those companies that operate in regulated industries or environments, you are going to see more strict adherence to the risk models that are enforced by leadership of the organisation. But organisations that managed risk were in line to “create nuanced opportunity for … everything from employee health and safety to revenue recognition."
Agentic AI in service desk implementation
One area Unisys has tapped into AI recently is in the embedding of agentic AI into its Service Experience Accelerator, a tool the company uses for service desk implementations.
The service desk solution claims to elevate agent productivity through automation, real-time translation, business insights and intelligent workflows, and is powered by the Service Experience Accelerator, harnessing governed generative AI capabilities.
“What we do is real time translation of inputs from 16 languages and counting - you can ask a question in any of these languages and get a response in any of these languages, so that's the traditional, Gen AI component of that,” Thomson said.
“The agentic AI component is the decision making that comes from the knowledge management database that sits underneath it that continually gets trained based on natural language input, and how that solves an issue, and then ultimately how it progresses level one and level two ticket resolution with a virtual agent.
“That's where you start to really see the compounding benefit of consistent answers, consistent learning processes, consistent delivery at low cost, where the knowledge can stay, from a data sovereignty perspective, in the tenant of the client and can start to cross functions.”
Unisys SVP and GM of digital workplace solutions, Patrycja Sobera, said today’s workforce does not work in one building, during standard hours or even speak one language, so outdated IT support models do not account for these changing dynamics, resulting in friction and bottlenecks that slow everyone down.
"Solutions (rather than tools) like Unisys Next-Generation Service Desk changes this by offering agent assist, touchless experiences and digital assistance, so employees can receive support whenever and however they need it," she said.
Sobera said the Service Experience Accelerator was developed to enable GenAI and machine learning integration as part of its core design from day one.
"While this framework has been fully rolled out and used by clients, it continues to be enhanced and expanded," she told techpartner.news.
"Some of the newer features – including the use of Agentic AI via our automation capabilities – are being piloted now and will be available next year."
This all ties into part of a broader trend that Thompson sees in-market around less hardware compute power and more refinement of the data model.
“Instead of a large language model, [why not] build a small language model that's fit for purpose and do it at the edge?” he said.
“You can do that type of computation at the edge with a model that's fit for purpose; you don't need the dynamic, large language model to crunch to a zillion variables that you're never going to use anyway.
“It comes back to master data management and knowing what you need the output to look like, how you're going to interface with that output, where the human in the loop is, and then ultimately, can you structure that in a manner that's dynamic and takes advantage of multiple inputs, but can be computed?”
Australia ‘outsized’ in level of importance
More locally, Thomson said the company is also seeing success with its work in the public sector, emphasising that Canberra is “probably [Unisys’] biggest hub in all of Asia Pacific and our biggest concentration of public sector clients”.
“We're scratching the surface on the amount of work that we can actually do in that region,” he said.
With APAC contributing nearly 15% of global revenue for Unisys, Thompson said Australia in general is outsized in its level of importance to the company.
“It is the lion's share of Asia Pacific from our perspective, and it has the most diverse client base,” he told techpartner.news.
“I think the things that are important to Australians, and specifically the public sector component of that market, are fully aligned to the solutions that we've developed and want to bring to bear there, and we've got some really long standing, multi-decade relationships in that region.
“[Australia] grew last year for us, and I expect it to grow again this year.”
.png&w=100&c=1&s=0)
