.jpg "It Still Feels Like Summer in Parts of the US. Blame El Niño")
Enterprise leaders are more sold on the promise of AI than employees, according to a panel of experts who also warned that the transformational benefits possible through adoption of the technology cannot be achieved if inefficient work processes stay the same.
Yang Li, chief operating officer and co-founder of the UK AI lab Cosine, said in a panel at Acceldata Autonomous 26 in London that he has been surprised to see “more buy-in from the top” than from the wider workforce, with employees showing “a significant amount of resistance” to using AI tools.
“We’ve spoken to a lot of S&P 100, FTSE100 companies where they’ve had Microsoft Copilot for years, they have AI-enabled tools, or they have AI native tools that they’ve rolled out across the organization. And we’ve seen adoption in single digit percentages,” Li said.
Instead, he argued, businesses are already seeing value in using AI agents to automate background tasks that humans are unlikely to complete, such as overnight cybersecurity testing.
“I think this is why I advocate to say, if you remove the human impetus out of it as an organizational transformation basis, you see a lot more acceleration,” he said.
Clare Schramm, managing director for business transformation and banking technology at Lloyds Banking Group, told attendees that leaders should focus on process design to create bespoke workflows for AI systems, rather than simply plugging them into existing human processes.
“You can’t just have your agents do the same process that a human does,” she said. “It will inevitably fail.”
Adoption for processes such as code generation and product development also has to coincide with business transformation, she explained.
“I think the biggest surprise for me has been when we started using AI to speed up prototyping phases, we thought that the business would absolutely love it because we’d be able to prototype really quickly,” Schramm said.
“But actually, the business was not ready for us to be able to prototype in an hour or in minutes. It actually forced them to think a bit quicker and also be ready for that, so that was one of the big surprises for us: we thought this would be this amazing value unlock and it was a very anticlimactic moment the first time that we vibe coded something.”
Ashwin Rajeeva, CTO and co-founder of Acceldata, told attendees that organisations cannot take too much time to put these changes in place, nor to figure out their AI strategy, or the technology will move on without them. The barrier here, he explained, is organisational rather than technical.
“No matter what we do as a technology provider, build AI factories or whatever, I think it’s harder to convince people to kind of use a new way of thinking around how they can move faster and adopt,” he added.
On the subject of driving change, Rajeeva noted that this is dependent on the use case. Programmers, he said, will naturally pick up a new tool and try it out in their work environments, while wider business processes take longer to transform.
“So you’ve got to get it top down and adopt quickly, and have enough people whether it is through [forward deployed engineers], people who create environments, or people who are specialised in running models on-prem.”
Schramm pointed to Lloyds’s chief executive Charlie Nunn as a leader that has been vocal and active in his support for AI, adding that that this provides her team of engineers with clarity and access to the tools they need.
At Lloyds, Schramm added, teams are already measuring AI-driven efficiency gains across areas such as customer relationship management. This demonstrates its value to build trust before introducing more automation for areas such as payment assistance and agentic payments.
“So we’re trying to lay some foundations, be very deliberate about use cases that we start to deploy, build up trust slowly over time, and gradually converge on larger value use cases.”
Li said that the most important thing organisations can do right now is try out new processes and attempt to test: “We definitely don’t know for certain if what we’re doing is on the right track, but sometimes just doing it is half the battle.”
