There’s a particular kind of energy that settles over a room when an industry stops pretending. At Gamma’s GX 2026, that moment arrived early – and it didn’t leave.
The event brought together enterprise IT leaders from across the full spectrum of UK industry: NHS trusts and local councils sitting alongside Europe’s largest retailers, manufacturers, utilities providers, logistics businesses, and financial institutions.
Different sectors, different pressures, different budgets. But when the conversation turned to AI, the same story kept surfacing. Strategy: strong. Execution: stalled.
That tension – between AI ambition and operational reality – defined the day.
The Pilot Purgatory Problem
If there was a single phrase that crystallised the mood at GX 2026, it was this: pilot purgatory.
Organisation after organisation has invested in AI experimentation. Proof of concepts have been built, use cases identified, demos delivered. And then – nothing. The pilot works. The scale doesn’t.
Alex Ayers, Sales Director at Gamma, spoke directly to this point.
“Most people are still treating AI as a technical experiment rather than an operational transformation,” he said.
“Pilots are working at a pilot stage, but they’re failing at the scale stage.”
It’s a diagnosis that resonated across the room, and one that cuts across sector boundaries.
Whether you’re an NHS trust navigating clinical governance or a retailer managing a sprawling legacy estate, the failure point tends to be the same: organisations are trying to deploy AI on top of processes, structures, and decision-making models that were never designed for it.
Redesign the Workflow, Not the Workforce
One of the sharpest interventions came when Ayers challenged the dominant narrative around AI and headcount.
The prevailing story – that AI’s primary value lies in replacing people – is not just reductive, he argued. It’s actively leading organisations in the wrong direction.
“You shouldn’t be looking at AI as a cost reduction exercise,” said Ayers. “It’s about organisational transformation.
“And to do that, you have to transform where work actually happens. Redesign the workflow, not the workforce.”
Organisations which had already moved to replace staff with AI agents were now looking to rehire – not because the technology had failed, but because the strategy had. Automation applied to a broken process doesn’t fix the process. It accelerates it.
There are, Ayers acknowledged, genuine efficiency gains to be had from automation. But chasing headcount savings as the primary objective sets a low ceiling – and distracts from the far larger prize of genuine operational transformation.
The System You Least Understand
Perhaps the most striking theme of GX 2026 was the one that had least to do with technology.
In conversations with enterprise leaders across Gamma’s customer base, Ayers found that the system most difficult to navigate in their AI programmes wasn’t data architecture or legacy infrastructure. It was their own organisation.”
“The system they least understood – the one they had to go back and revisit – was themselves,” he reflected.
Data, governance, process design: these were all cited. But underpinning all of it was something more fundamental – the challenge of making good decisions across complex, multi-stakeholder organisations.
“When you layer in a disruptive technology that has the power to drive true business value, you end up with the perfect storm,” Ayers said.
“Something that can easily go right – but probably more often than not, can easily go wrong.”
Governance and the Road to Agentic
Conversations have shifted from present challenges to future stakes. Multi-agent systems, autonomous AI, and what Ayers described as “agent factories” are no longer theoretical. They are the next horizon – and the distance between organisations that are ready and those that are not widening quickly.
The governance conversation became central here. Shadow AI – departments independently adopting AI tools outside any overarching strategy – is emerging as the enterprise equivalent of shadow IT.
Leaders across the room described teams unsure which AI tool to use, inconsistent governance models, and accountability gaps that no one had quite mapped yet.
The message is clear: if you can’t deliver one pilot well – end-to-end, governed, secure – you are not ready for what comes next. And what comes next is arriving faster than most organisations appreciate.
The North Star
GX 2026 didn’t offer easy answers – but it did offer a clear direction.
The organisations making genuine progress with AI are not the ones running the most pilots or deploying the most tools.
They are the ones doing one thing brilliantly – engineering it properly, end-to-end, as a business transformation initiative rather than a technology project.
“We don’t have an ideas problem with AI,” Ayers said. “We have an execution problem.”
For the enterprise IT leaders in the room – from the NHS to the high street, from utilities to local government – that framing felt less like a provocation and more like a relief. The path forward isn’t about doing more. It’s about doing less, better. Picking one use case, committing to it fully, and building the organisational muscle to scale from there.
