Your next hire might not have a desk – or a heartbeat. What once belonged to the realm of science fiction is now rapidly becoming a reality in business strategies and boardrooms. A new class of artificial intelligence, known as agentic AI, is capable of planning, acting and adapting with minimal human supervision. As businesses strive for better efficiency and agility, this emerging technology presents both opportunities and challenges.
To explore how enterprises can safely implement agentic AI at scale and why there is an urgency to do so, Raconteur assembled a panel of experts for a timely webinar discussion.
A truly transformative technology
New technologies typically improve how businesses operate. The internet enabled better connectivity. The cloud moved software and data to a more flexible space. But agentic AI will usher in a more fundamental shift in business operations.
This technology means we have to truly look at business processes and the way we work
That was the view of Robert Whitaker, vice-president and CIO for industry go-to-market, UKI and APAC at Salesforce. He said: “This technology means we have to truly look at business processes and the way we work. And that’s going to require real transformation, rather than just a shift in technologies.”
Maxime Vermeir, director of AI strategy at Abbyy, agreed: “Many organisations will explore this technology and find the technology points back at them to figure out if they know how the organisation works.” He explained that this isn’t just about having data in order, but it’s also about understanding how processes are meant to function and how people execute them.
Why is there such a sense of urgency around agentic AI, despite the term having been in the business lexicon for less than a year? According to Cameron Williams, head of technology architecture at Acre Pharmacy, it’s because AI is now ubiquitous. “We have AI in every aspect of life now,” he said. “People expect that service and quick turnaround. They expect someone to be able to deal with their problems instantly, and if you aren’t providing that to them, they will find someone who will.”
Agents in action
Acre is an example of a company early in an ambitious agentic AI journey. Williams explained that the company’s first agentic AI use case was a customer service agent. It might look like a regular chatbot on the surface, but it can serve the customer without getting a human involved. He explained that their new AI agent can request prescriptions or update deliveries, saving the customer service team valuable time.
“We’ve actually changed our customer service team to a combined customer and account management team, because they’ve now got that extra time available from dealing with queries to make sure that customers are happy at the account level,” he said. “We’re seeing 55% fewer calls into the office, but we’ve doubled the business in that time frame.”
Many organisations will still have reservations about the safety of this new technology. Williams acknowledged this and emphasised that, as an online pharmacy, it’s essential that data is kept secure, clean and used ethically. “We have a responsible person put their name against every decision that’s made to make sure that it’s being checked thoroughly.”
He also highlighted the importance of platform choice. “The best thing about Agentforce is the way it anonymises all the data before it goes into the model. It lives within our organisation, and the profiles and permissions we have set up are all safe. That removed our worries about the ethical side, but it’s something that you’ve always got to think about, especially in our industry.”
Whitaker added: “At Salesforce, we do everything based on our core values of trust and equality. Customers are not going to want their data to go into models and be used elsewhere. So we put a lot of effort into building those values into the platform.”
He shared a telling anecdote: a customer’s compliance officer initially rejected any use of AI in the organisation. “After we talked through the controls built into the platform, he said, ‘Why would I ever let a person talk to a customer without having AI check the answer?’. He realised those controls are actually stronger than the human controls and changed his mind. That’s the scale of the shift.”
The evolution of ‘human-in-the-loop’
While the early assumption was that humans would always remain ‘in the loop’ when working with AI, the discussion revealed that this dynamic is shifting in some cases.
Vermeir explained that Abbyy has always maintained a human-in-the-loop approach. The company’s focus is on how organisations can leverage the unstructured data that’s locked away inside their documents, and this unstructured information has to be validated before it ends up in the ERP system, CRM or system of record.
“That’s always been very important,” he said. “We’ve always tried to optimise that so the manual effort required on the customer side is reduced to a bare minimum. And I think customers will still be looking for that level of freedom.”
We are starting to see that AI can take action without a human in the loop, particularly in those customer service scenarios
He suggested that the level of human involvement will vary by industry and organisation, and that it’s crucial to assess which processes truly require manual verification and which can be fully automated. “It’s striking that balance between which process is right for not just agentic AI, but any AI or technology. That is an exercise organisations need to do before they start on their journey for each process.”
For regulatory or high-stakes processes, AI may do the legwork, with a human providing the final sign-off. For routine business processes, agentic AI is increasingly taking full action without human involvement, as long as robust guardrails are in place.
“We are starting to see that AI can take action without a human in the loop, particularly in those customer service scenarios,” explained Whitaker. “If I just look at how we use it on our customer support site, we’ve now done about 750,000 conversations with customers through our AI without a human intervention in that process – and 85% of those conversations have been resolved without a human ever touching or being handed off to those cases. So you’re seeing more of these examples where people are getting comfortable with AI.”
Eyes on the road ahead
The robot revolution isn’t slowing down, so leaders need to keep their eyes on the road ahead. The panellists shared their thoughts on what leaders need to be aware of as the technology evolves, as well as some exciting developments on the horizon.
Vermeir noted that the development of robust trust and compliance frameworks is critically important for future success and continuous innovation, and that getting ahead of regulation will be key.
Whitaker flagged cybersecurity as a growing concern. As with all rapidly evolving technologies, AI can be exploited by bad actors. Deepfakes that can replicate voices and bypass traditional security controls pose a real threat, and detection methods will need to advance quickly.
On a more optimistic note, the panel was excited by the prospect of voice-enabled AI capable of having conversations rather than responding in text. There could be a future where AI answers inbound calls and even performs outbound sales calls.
Vermeir concluded with his utopian vision of the future. “I’m looking forward to the point where Star Trek becomes a reality and we will be able to simply ask things and have the computer execute them.”
Your next hire might not have a desk – or a heartbeat. What once belonged to the realm of science fiction is now rapidly becoming a reality in business strategies and boardrooms. A new class of artificial intelligence, known as agentic AI, is capable of planning, acting and adapting with minimal human supervision. As businesses strive for better efficiency and agility, this emerging technology presents both opportunities and challenges.
To explore how enterprises can safely implement agentic AI at scale and why there is an urgency to do so, Raconteur assembled a panel of experts for a timely webinar discussion.
A truly transformative technology
New technologies typically improve how businesses operate. The internet enabled better connectivity. The cloud moved software and data to a more flexible space. But agentic AI will usher in a more fundamental shift in business operations.
