From ~8 direct reports to 15+
From manager to player-coach.
From human teams to one-person pods with agents
That's the org structure @brian_armstrong outlined in his @coinbase memo today. It came alongside the announcement of a 14% layoff...the latest in a string of tech layoffs this year citing AI as a primary reason.
In describing how Coinbase will operate going forward, three things stand out:
1. Fewer layers, faster decisions: A flattening of the org structure to 5 layers max below the C-suite. Leaders will own more, and span of control will increase to 15+ direct reports
2. No pure managers: Every manager has to be be a player-coach; a strong, active individual contributor alongside their team
3. AI-native pods: Concentrating around AI-native talent who can manage fleets of agents and experimenting with “one person teams” with engineers, designers, and product managers all in one role.
For decades, the manager career path was to stop doing and start managing. AI is now inverting that.
What's not being said out loud:
AI can handle the coordination layer of scheduling, status updates, and pulling context across tools. That's what makes 15+ directs theoretically possible.
But coaching, giving feedback, and decision-making are a series of human judgment calls and difficult conversations that require both time and context.
If managers offload coordination to agents and reinvest that time into the human side of the job, the model may work. But the tools managers have today were built for the 8-direct-report world. They don't scale, and the org chart is being redesigned faster than the manager toolkit can keep up.
That's why we're building @trywindmill: a context graph for your people. It's a continuous picture of every person on your team from what they're working on to who they're collaborating with, and how they're doing. And it's all pulled from where work already lives.
A transition like this isn't possible without real-time visibility into how your people are adjusting to new tools, new expectations, and roles being redesigned constantly.
For decades, span of control was capped by what humans could manually observe. AI is lifting the ceiling, but only for companies that invest time and tooling into actually seeing their people.