Agentic artificial intelligence (AI) systems, comprising multiple agents working collaboratively, will drastically transform business performance in the coming years, according to an industry expert.

Speaking on an episode of GlobalData’s Instant Insights podcast, Chris Ashley, vice president for strategy at Peak, which optimises business systems using AI, predicted that agentic systems mapping back to complex business processes will provide “enormous uplift in business performance” in the next couple of years.

“Rather than just leveraging general-purpose agents like ChatGPT’s very impressive agent, we’ll start to see systems that have 10, 20, 30 agents all interacting together and collaborating together to work on really, really complex tasks,” Ashley predicted. “I’m personally really excited about that. I think this is where we’ll see breakthroughs in science, in physics, on some of the world’s hardest problems over the coming years.”

Ashley was appearing on the podcast following OpenAI’s launch of the ChatGPT agent, a consumer-facing implementation of agentic AI that can carry out complex tasks using its own reasoning. Agentic AI has been emerging as the next major iteration of the wider technology, with GlobalData analysis outlining that businesses are already “pivoting from building deterministic workflows to implementing AI agents that learn, adapt, make decisions and perform complex tasks.” As with its launch of ChatGPT for generative AI, OpenAI’s launch of the ChatGPT agentic is likely to significantly raise the profile of agentic AI in the public consciousness.

“These general-purpose agents have captured the consumer imagination and the zeitgeist AI moment,” said Ashley. “I think, where I’m definitely interested is in that as a consumer, in these agents and the possibilities, I also really like to look at the specialised agents as well. I think they have the capacity to help us achieve really meaningful uplifts within business environments.”

Agentic AI use cases

Among the use cases Ashley sees for businesses are quote pricing, markdown and promotions, sales and operations planning, scenario planning within the supply chain and merchandising.

“I think that the core benefits are all focused around enhanced speed of decision making, enhanced scalability of decision making and compound returns on investment,” he explained. “What I mean by that is we’re working with manufacturers, consumer packaged goods companies and retailers around the world, and we’re seeing specialised agentic capabilities being deployed now. This can be within a manufacturing environment, agentic systems for quote pricing – so businesses that respond to really large volumes of quote requests from a customer base, often a complex basket of goods.

These processes, a couple of years ago, would have taken seven, eight hours to complete, been really burdensome to grab all of the structured and unstructured data, analyse it, process it, package it into a quote response, send it back to a customer.

“We’re now seeing agentic systems being leveraged, where you can respond within minutes, and the compound impact on that is obviously enormous. You’re returning many, many tens of thousands of hours of time back into a commercial sales team that can focus on building relationships and grabbing more market share, rather than having heads in spreadsheets analysing complex data and responding to quote requests.”

Risks of agentic AI

Despite the huge potential benefits for businesses of agentic AI, Ashley also noted that there are risks that must be taken into account.

“You’re leveraging software systems that now have levels of autonomy that we’ve never witnessed before within software, because they’re no longer hard-coded and deterministic. So, I think that is both the major feature, but also can be a risk, inevitably … “If you have poorly mapped processes within your business, if you have a lack of standardised data governance, if you have a lack of a governance model for the AI agents themselves to monitor the performance of those agents, then you’re exposing yourself to heightened levels of risk by deploying these systems, because they can obviously compound.”