You don't have an AI Adoption Problem, you have an AI Collaboration Problem

I was at the House of Lords yesterday for CambrianEdge.ai's UK launch and the release of their first research report, AI at Work: The Collaboration Gap. Not a bad venue to discuss the future of work as venues go. The room was a genuine mix of finance, academia, technology, philanthropy and enterprise leadership, all grappling with the same question: why is AI not delivering what we expected?

The answer, backed by data from 775 users across 104 organisations, is not what most boardrooms want to hear. The problem is not the technology. The problem is that AI is still a solo sport.

The collaboration gap

55% of professionals identify solo use or the absence of any structured human-AI workflow as their biggest challenge. 62% of organisations have no defined process for handing off AI-generated work to human review. 27% have zero collaboration infrastructure. No shared access, no prompt libraries, no training, no quality standards.

The bottleneck is not adoption. It is the inability to move from individual experimentation to structured team workflows where humans and AI operate as a continuous system. Harjiv Singh, CambrianEdge's founder & CEO, put it well: the tools arrive first. Transformation follows, but only when organisations redesign the work itself, not just add new instruments to old workflows.

When the infrastructure is not there, the consequences are real. 18% of organisations have already rolled back or abandoned AI initiatives entirely, citing quality collapses and adoption failures. The dividing line is infrastructure, not ambition.

Drucker's knowledge workers meet AI

Being in the room yesterday made me reflect on the parallels with Peter Drucker’s term "knowledge worker" coined in 1959. His argument was that the primary capital of a modern organisation is not its machinery or its processes but its people: the professionals who think, solve complex problems and drive innovation. He later said knowledge worker productivity would be the next frontier of management.

Sixty-seven years later, we have a tool that can genuinely augment every knowledge worker in an organisation. Yet we are currently treating it like a personal productivity hack rather than an organisational capability. The connective tissue that turns individual use into collective intelligence simply does not exist in most businesses. Knowledge capital is as critical as data and it disappears when an individual leaves or is not enabled to share what they have learned.

The electricity analogy to reflect the current state of AI used in the room was also apt. In the 1920s, some companies were still lighting candles. Others were redesigning entire cities. The infrastructure gradient determined who won. We are at exactly that point with AI.

Why leaders should be paying attention

This is not just a technology leadership problem and I don’t envy any CIO at the moment. BCG's 2026 survey of 300 global CMOs found that 96% claim AI is driving end-to-end transformation of their marketing function. Yet 42% still use AI only to assist humans with discrete individual tasks. BCG calls this the "transformation illusion." If it is happening in marketing, one of the most AI-exposed functions, it is happening everywhere.

The triangle of people, process and technology is one every marketing leader has to keep spinning. Experimenting with AI has ticked a visible box on the technology side. But like martech before it, the real value sits in the process layer: the workflows, the handoffs, the quality standards, the collaboration infrastructure that turns experimentation into execution. Without that, you have tools generating output that nobody is reviewing, sharing or building on.

For PE-backed businesses running to a value creation timeline, this matters more. You cannot afford 12 months of AI experimentation that produces activity but not impact. The leaders who will create real competitive advantage are the ones who can diagnose where AI fits into the commercial engine, build the collaboration infrastructure around it and make it a team capability rather than a collection of individual workarounds. The challenge is having the time to do this alongside the day job.

Collaborative infrastructure is the strategy

There is a version of the next two years where AI becomes background noise. Another technology that promised transformation and delivered incremental efficiency. The companies that treat collaboration infrastructure as seriously as they treat the tools themselves will pull ahead. Not because they adopted faster, but because they built the architecture for their people to actually use what they adopted.

Lord Loomba's opening words at the event stayed with me: as we shape the future of AI, intelligence remains human. The collaboration, the judgement, the standards. These are ours to build.