When something goes wrong in your organisation, how many people can honestly say they own the outcome? Not attended the meeting. Not contributed to the discussion. Owned it. If the answer is nobody, you have a diffusion problem that is costing you speed and accountability.
Apple popularised a management concept decades ago called the Directly Responsible Individual, or DRI. The idea is simple. Every project, initiative, or activity has exactly one person who is ultimately accountable for its success or failure. Not a committee. Not a team. One human being who knows that when something goes wrong, the buck stops with them.
GitLab documented this concept in its handbook, and it is one of the clearest explanations available. The DRI is not necessarily the person doing all the work. They are the person who ensures the work gets done. They coordinate, they escalate, they make the hard calls. Everyone else knows who to go to with questions and who to hold accountable for results.
Why diffusion kills delivery
The alternative is shared ownership, which in practice means no ownership. When three people are responsible for a delivery, each one assumes someone else is driving. Status meetings generate updates but nobody unblocks. Decisions get deferred because no single person has the authority to make them.
Research on the Ringelmann effect has shown for decades that individual effort declines as group size increases. People pull less when they believe others will pick up the slack. A DRI eliminates that ambiguity. The message is not that the team does not matter. It is that someone needs to be answerable for the outcome. Teams with clear DRIs move faster because decisions happen at the right level rather than being escalated to managers who lack context.
What a good DRI looks like
A good DRI has three things: clarity of scope, decision authority, and a real stake in the outcome. They know exactly what they own. They can make decisions without running everything up the chain. And they feel the consequences when things go wrong.
This is not about blame. It is about accountability. A DRI who hits a blocker escalates it immediately because they own the timeline. They do not wait for the weekly standup to mention that a dependency is late. They track it, they push it, they find a workaround. When a DRI needs help, they ask for it directly because there is no ambiguity about who is responsible for the outcome.
The trap: AI agents cannot be DRIs
As AI agents become more capable, there is a growing temptation to treat them as accountable for work outcomes. A coding agent that ships features. A support agent that resolves tickets. A sales agent that books meetings. Should the agent be the DRI?
The answer is no, and the logic is as old as computing itself. IBM made this point in a legendary 1979 training slide: a computer can never be held accountable, therefore a computer must never make a management decision. An AI agent cannot take responsibility. It cannot be fired. It cannot feel the consequences of failure. Accountability is a uniquely human concept.
The agent can be the tool. The human who deployed it and who oversees its output is the DRI. That distinction matters when something goes wrong and you need to know who answers for it. Never delegate accountability to a system that cannot be held accountable.
Putting DRI into practice
Start with one project. Assign a single DRI. Give them the authority they need. Tell everyone else that questions go to that person. When the project finishes, debrief on whether the clarity helped. Most teams find that the DRI model surfaces problems faster, reduces meeting overhead, and produces better outcomes. It also reveals who your strongest operators are. The people who thrive as DRIs are the ones who want ownership, not visibility. Try it on your next initiative. Pick one person. Make them accountable. See what happens.