Systems Don't Care About Your Org Chart
Organizations write characters on people. When AI enters org dynamics, it responds to actual information flows — not the hierarchy.
There’s a phrase I’ve been sitting with for a while: organisations write characters on people.
I first heard it framed by Mark Claydon, and it’s stuck. The idea is that organisational structures — the hierarchy, the incentive systems, the information flows, the cultural norms — don’t just affect how people behave. They shape who people become inside that system. The ambitious person in a siloed org becomes hoarding and territorial. The same person in a genuinely collaborative structure becomes generous. Same person. Different character, written by the org.
Systems thinking says something similar. The system produces the behaviour. If you want to change the behaviour, change the system.
Here’s what I keep noticing about AI in organisations: it doesn’t respect the org chart.
The org chart is a formal document about who reports to whom. It’s a representation of authority, accountability, and (in theory) information flow. But the actual system — the one that produces outcomes — is the network of real information flows. Who actually talks to whom. Whose emails get read. Which Slack channels contain the real decisions, and which contain the announced decisions.
AI, when it enters an organisation, responds to the actual information flows. Not the formal ones.
If a team’s knowledge is fragmented across inboxes, AI will surface that fragmentation. If decision-making authority is formally delegated but practically centralised, AI-assisted analysis will keep routing back to the centre because that’s where the validated context lives. If the org chart says “product and engineering are peers” but product people have never been taught to write specs that engineers can use, an AI coding assistant makes that gap bigger, not smaller — it just generates faster.
This matters for org design in ways I don’t think we’ve fully reckoned with.
For the last couple of years, the conversation has mostly been “which jobs will AI replace?” That’s the wrong level of analysis. The more interesting question is: what does AI amplify? Because it tends to amplify what’s already there. Effective teams with good information hygiene get more effective. Dysfunctional teams with unclear accountability get more confused faster.
AI is a structural revealer. It shows you what the system actually is, not what the org chart says it is.
I’ve sat in enough workshops to know that this can be threatening. Nobody likes being told that the thing they put on the slide about “empowered, cross-functional teams” is contradicted by the actual communication patterns in their Slack workspace.
But it’s also an opportunity. If you’re willing to look at what AI’s presence in your org is actually surfacing — where it’s producing confusion, where it’s bypassing the formal structure, where it’s being quietly avoided — you have a diagnostic that most consultants can’t give you.
The org is writing a character on your AI too. Worth reading what it says.