Here’s a thought that’s stuck with me following my recent SAATKORN podcast conversation with Sebastian Walker, the CEO and founder of Omnora: Imagine you were to hire an intern with an IQ of 250. This person can analyse almost anything, speaks every language, works round the clock and has read just about everything there is to read.
Sounds like a dream candidate.
There’s just one problem: on Monday morning, this super-intern knows absolutely nothing about your company. And that, I think, sums up pretty well where we currently stand with enterprise AI.
The missing piece of the jigsaw is context
We spend a lot of time discussing which model is more intelligent, which agent can handle which task, and how quickly AI capabilities are improving. Impressive stuff. But the intelligence of the model itself will become less and less of a distinguishing feature. Because anyone can buy it.
What you can’t buy is your organisation’s collective experience. Why a particular process works the way it does. Why that one customer needs to be treated differently. Why a project failed five years ago. Who knows whom. Which official rule everyone ignores for very good reasons.
Every organisation has a Claudia from the finance department who knows how things really work. Unfortunately, Claudia isn’t in the database.
We are facing a lack of context
At present, two developments are affecting businesses simultaneously:
- AI is evolving from the experimental stage towards practical application. Microsoft’s ‘2025 Work Trend Index’ found that 82 per cent of executives expect to deploy digital workers to expand their workforce within the next 12 to 18 months.
- At the same time, experienced staff are leaving organisations, whilst demographic change is accelerating. We usually refer to this as a skills shortage. But I believe we need a different term for what is currently happening: a lack of context.
Because when experienced employees leave a company, they don’t just take their skills with them. They take their decisions, experiences, relationships, failures, shortcuts and unwritten rules with them. This is precisely the sort of knowledge an AI agent needs if we expect it to do more than just summarise documents and write emails.
Sebastian Walker and I discussed this in detail on the podcast: organisational knowledge isn’t just what you find in SharePoint, manuals or process documentation. A huge part of it is stored in people’s heads. And as soon as those people leave the building, good luck asking SharePoint for help.
Integrating the organisation into AI
This is changing my understanding of AI maturity. Installing yet another Copilot, LLM or agent is no longer the difficult part. The interesting question is whether we can provide these systems with the organisational context they need to actually create added value.
McKinsey’s 2025 AI study found that, among the organisational factors examined, the redesign of workflows showed the strongest correlation with the impact of GenAI on EBIT. Nevertheless, only 21 per cent of respondents using GenAI stated that their organisations had fundamentally redesigned at least some of their workflows. It is therefore not enough simply to overlay old structures with powerful technology.
The challenge is not bringing AI into the organisation. The challenge lies in integrating the organisation into AI.
And this is precisely where the HR department needs to pay close attention. If future teams consist of people and AI agents, someone must consider how both sides acquire the knowledge they need for their work. IT naturally plays a crucial role in this. But this is not purely an IT issue.
The HR department is well versed in onboarding, training, skills, succession planning and organisational development. And it should know where critical knowledge is located – especially when the people who hold that knowledge are about to leave the company.
This creates a much more interesting role for HR in the AI transformation: not just preparing people for AI, but preparing organisational knowledge for both
people and
AI. Much more exciting than the next prompt training session, if you ask me.
MY OPINION
Over the last two years, we’ve been obsessed with making AI smarter. Understandably so. But the models are slowly becoming intelligent enough. The real question is whether they understand our organisations.
Your competitors can buy access to the same AI models. But they can’t buy your experience, your customer knowledge, your decisions, your failures and everything your staff have learnt over decades. Perhaps that’s one of the real competitive advantages in the next phase of AI. So yes: give me the intern with an IQ of 250.
But please do train the bloody intern first before you let him run the company. 😉
LISTENING TIP
The idea for this column came from my recent SAATKORN podcast conversation with Sebastian Walker, CEO and founder of Omnora
. We discussed why traditional knowledge management falls short, how tacit knowledge can be captured, what AI agents need to become useful in organisations – and why the HR department plays a bigger role in this than many CHROs might currently realise. Listen here: