The Most Experienced Person in the Room Might Be AI

Why the future of AI may be less about replacing expertise and more about sharing it.

For years, organisations have invested heavily in capturing knowledge. Policies have been written, procedures documented, training courses designed and knowledge bases built in the hope that expertise could be shared more widely. Yet despite all of that effort, almost every organisation still relies on a handful of people who simply know how things work. They know which regulation changed six months ago, remember why a particular exception exists and can often spot problems long before anyone else notices them. Their expertise hasn’t usually come from reading manuals. It has been built through years of experience, conversations, mistakes and judgement.

The problem is that experience doesn’t scale very well. Experienced people retire, move on or simply can’t answer every question from every colleague. Documentation helps, but it often removes the context that made the knowledge valuable in the first place. A policy might explain what to do, but it rarely explains why the rule exists, when it should be applied differently or what has happened in similar situations before. Organisations have become very good at storing knowledge, but much less successful at making it available at exactly the moment people need it.

During a discussion with Brian Feeney from AI-XON, I found myself thinking about this challenge in a completely different way. Rather than describing AI as something that automates work or replaces people, Brian talked about building systems that make organisational expertise available to everyone. It was a subtle shift in language, but one that completely changed how I thought about workplace AI.

Learning Before the Mistake Happens

Brian explained that the idea emerged while working in the net zero sector. The engineers and installers weren’t struggling because they lacked technical ability. They were highly skilled at what they did. The challenge came from navigating an increasingly complex landscape of compliance requirements, evolving standards and local variations. Even experienced people could miss small details, not because they were incapable, but because they were expected to hold an enormous amount of constantly changing information in their heads while completing their work.

Most organisations deal with that problem in a familiar way. Work is completed first and reviewed afterwards. Audits identify mistakes, quality assurance highlights missed steps and compliance teams explain what should have happened. That approach has worked for decades, but it also means learning often happens after the cost has already been incurred. By the time someone discovers a missing document, an incorrect process or an overlooked regulation, the work has already been completed and the opportunity to avoid the mistake has passed.

Rather than improving the audit itself, Brian described asking a different question. What if the knowledge sitting behind that audit could be moved to the beginning of the process? Instead of telling people what they should have done yesterday, could AI help them make better decisions while the work was actually taking place? That question eventually became Guardian.

A Different Kind of AI

Listening to Brian, I realised this wasn’t really a story about chatbots or virtual assistants. Those technologies are useful, but they tend to focus on completing tasks more quickly. What he was describing felt fundamentally different. It was an attempt to make years of organisational experience available to someone carrying out their job, regardless of whether they had been with the organisation for three months or thirty years.

Imagine being able to ask questions of a colleague who never forgot a regulation, remembered every previous case, understood every internal policy and stayed completely up to date with changing standards. That colleague wouldn’t replace your judgement, but they would almost certainly help you make better decisions. They would reduce uncertainty, provide context and help you avoid mistakes before they happened. What Brian was describing wasn’t simply AI answering questions. It was AI helping organisations distribute expertise far more effectively than they have been able to before.

That feels like a much bigger idea than workplace productivity. If organisations can genuinely make their best knowledge available to everyone, experience itself begins to scale in a way that has never really been possible before.

Rethinking Experience

Much of the public discussion around AI still revolves around replacement. Which jobs will disappear? Which tasks can be automated? Those are important questions, but they risk overlooking another possibility. AI may prove just as valuable because of what it enables people to do, not simply because of what it does instead of them.

The organisations that benefit most from AI may not be the ones that remove the greatest number of people from a process. They may be the ones that become much better at sharing what their most experienced people already know. In that world, AI doesn’t replace expertise. It extends it, making decades of accumulated knowledge available wherever and whenever it is needed.

Perhaps that’s a more interesting way to think about the future of work. Not that AI becomes the employee. But that every employee gains access to the experience of the very best colleague they’ve ever worked with.