Joining me on the panel were Neil Pragnell, Head of Portfolio Delivery at NHS Business Services Authority, Ross Mardell, Chief Product Officer and Co-Founder at Verto, and Natasha Osborne, Head of Client Success and Change Practitioner at Verto. Together, we explored what is preventing organisations from moving beyond experimentation and towards genuine AI adoption.
What emerged was a surprisingly human answer to what many assume is a technology problem.
The starting point for our discussion was simple. AI tools have become commonplace for activities such as writing, research and automating routine tasks. Yet very few organisations are using AI to improve programme delivery, strategic decision making or portfolio management at scale.
According to Neil, the reason is that individual adoption is easy.
"The barriers to adoption for an individual are way lower. You need to log in and a device, which we all have. So, as an individual, it's very easy to maximise AI and use it every day."
The challenge begins when organisations attempt to operationalise AI. Neil highlighted the reality facing public sector leaders:
"It becomes far more complex. There's a lot more regulatory aspects. There's policy guidelines, GDPR to consider, standards, ethics, all of these things."
This reflected one of the key themes built into our panel preparation: organisational adoption is being held back not by technology alone, but by governance, trust, change management and data quality.
One of the most interesting moments came when I asked Ross whether the technology itself is mature enough to support large-scale portfolio analysis. His answer was clear.
"The technology is certainly ready to do it."
However, he quickly added several caveats around governance, security, privacy and data quality. As Ross put it:
"If you put rubbish in, you'll get rubbish out. However, in the AI world... you put rubbish in, but you get sensible rubbish out."
That observation resonated strongly with the audience. AI can generate outputs that appear entirely plausible, even when the underlying data is poor. This makes data quality more important than ever.
Ross argued that technology is no longer the primary barrier: "I think the tech is ready, but the humans aren't." And that statement became a recurring theme throughout the discussion.
Natasha took the conversation further by focusing on the human side of adoption. Her view was straightforward: "It's absolutely a people problem."
She argued that organisations have largely taken a reactive approach to AI, rather than treating it as a structured business change programme: "AI has moved so fast that what we've been doing this whole time is a reactive approach instead of a proactive approach."
People worry about accountability. They worry about decision-making. They worry whether AI will change or even replace aspects of their role. These concerns were also identified in our panel planning notes as one of the most significant barriers to adoption.
Natasha challenged organisations to rethink how they communicate the purpose of AI: "What work can humans stop doing that AI can take over so that humans can focus on higher value activities?"
It is a simple question, but one many organisations have yet to answer.
The second part of the discussion focused on what happens when AI moves beyond basic productivity tools and becomes an active participant in programme delivery.
Verto has approached this challenge through the use of specialist AI agents. Instead of creating one generic assistant, Verto has developed agents with specific responsibilities. As Ross explained: "We've actually got around about 18 agents out-of-the-box in our system."
Examples include:
Each has a defined role and area of expertise, and people naturally begin to interact with these agents as if they were members of the team.
"I'm saying her or him, because I'm starting to relate to them as a person" explains Ross.
Interestingly, this was where Natasha identified another adoption challenge. While personification makes AI easier to understand, it can also make it more intimidating. "For a lot of people, this actually makes it seem a bit scarier too."
Her advice was a valid concern. However, thoughtful change management assures the human team: "The AI isn't taking away your job. It's freeing you up to be able to do more with the time that you have." Organisations must actively communicate that AI exists to elevate people rather than replace them.
The panel's most memorable phrase belonged to Neil.
Discussing the dangers of unchecked AI usage, he warned about what he called "AI sloppification." His concern was that if AI generates reports, AI reviews those reports, and AI then summarises those reports for decision-makers, organisations risk losing genuine understanding of what is actually happening. Neil put it perfectly: "It's a supplementation to the human intelligence and not an outsourcing of human intelligence."
That distinction matters enormously. The future is not AI replacing professional judgement. The future is AI supporting better professional judgement.
The final section of the panel focused on practical actions. The Programme and Project Data Standard featured prominently throughout the conversation and was highlighted as a foundational requirement for organisations that want to become AI-ready. Trusted, standardised project data provides the basis for effective AI adoption.
Neil's advice centred around three fundamentals:
Ross emphasised the importance of starting small: "Start with a pretty narrow use case."
Organisations should begin with controlled datasets and clearly understood problems, building trust and confidence before expanding further.
As Ross explained: "You start to trust it because you're trusting that it's giving you the correct output based on known data."
Perhaps the most important lesson from the session is that AI adoption is not primarily a technology programme. It is a data programme. It is a governance programme. Most importantly, it is a people programme.
As Natasha reminded us, organisations need to create environments where people understand how AI helps them succeed, rather than fearing what it might change. And as Neil concluded: "Don't assume AI is the right solution, but always consider it as one flavour of solutions on the table."
The adoption gap is real. But it is not insurmountable.
The organisations that succeed will be those that focus less on the latest AI capability and more on the fundamentals: quality data, clear governance, structured change management and empowered people.
Only then will AI move from being an interesting tool to becoming a genuine strategic advantage.