For the past year, most conversations about AI have focused on basic productivity—like helping developers write code a bit faster or adding a "copilot" to help with daily tasks. While those tools are useful, they only scratch the surface of what is actually changing.
The real shift isn't just about individual speed; it is about fundamentally changing how we organize our teams and work together. To understand why this is a big deal, it helps to look at why our companies are structured the way they are in the first place.
The 2,000-Year-Old Coordination Problem
Every large organization faces the same basic problem: how do you coordinate a lot of people to get something done?
The Romans were among the first to solve this on a massive scale. Through centuries of running an army, they discovered a basic human limitation: one leader can really only manage about three to eight people effectively. To coordinate an army of 5,000, they built a nested hierarchy. Eight soldiers reported to one leader, ten of those groups reported to the next level up, and so on.
Later, the Prussian army took this a step further by creating the "General Staff"—essentially the first middle managers. Their entire job was to process information, plan, and route messages up and down the chain. By the 1850s, the American railroads adopted this exact military structure to manage thousands of workers and prevent train collisions, giving us the very first corporate org charts.
For over a century, we've relied on this same structure. Even modern tech companies that have tried to run "flat" organizations usually end up reverting to a traditional hierarchy as they grow. Why? Because narrowing the "span of control" means adding layers of management, and those layers have always been the only reliable way to route information and keep everyone aligned.
Until now.
AI as the Coordination Layer
If we step back, a lot of traditional middle management is essentially human middleware—people routing information, summarizing status updates, and trying to keep teams on the same page.
Today, AI can take on that coordination role. Instead of treating AI as just a sidekick for writing emails, we can use it as the underlying operating system for our teams. This allows us to shift from "open loops" to "closed loops."
- Open Loops (The Old Way): Decisions are made, people execute them, but the lessons and context get lost in private chat threads, forgotten documents, or siloed departments. The information flow is leaky.
- Closed Loops (The AI Way): The system continuously monitors the output of a project, learns from what is actually happening in the business, and automatically helps you adjust the next step.
Rethinking Status Updates and Management
Let's look at how this changes a common team workflow: sprint planning.
In a traditional setup, status updates often feel like a game of telephone. A manager asks a team lead how a project is going, the lead asks the developers, and everyone tries to remember what they did over the last two weeks. It is slow, manual, and often inaccurate.
The Core Shift: Instead of relying on people to manually route, summarize, and pass information up and down the chain, we can let an intelligence layer handle the coordination.
When a company's data is readable, an AI agent can look directly at your actual work—like task trackers, code commits, customer feedback, and meeting notes. It doesn't guess or rely on vague summaries; it looks at the raw data to see what actually shipped and what didn't. This helps teams plan their next steps with much higher accuracy.
| Area | Traditional Approach | The Collaborative AI Flow |
|---|---|---|
| Information Flow | Filtered through layers of meetings and messages | Direct access to raw, documented team context |
| Status Updates | Manual rollups, status meetings, and guesswork | Automated, data-backed progress tracking |
| Coordination Cost | High (requires constant meetings to align) | Low (AI handles routing and documentation) |
Making the Company "Queryable"
For an AI system to help coordinate your team, it needs to be able to see what's happening. Your organization has to become legible, or "queryable."
This means making sure our daily work leaves a digital footprint that an AI can learn from. If your team's context is locked away in private, fleeting direct messages or unspoken thoughts, the AI cannot help you. We can fix this with a few straightforward habits:
- Documenting conversations: Using AI to record and summarize meetings.
- Working in the open: Moving away from isolated DMs and keeping project discussions in shared, centralized channels.
- Creating a "World Model": Connecting your task trackers, code repositories, and customer feedback into a single source of truth that the AI can read.
When you do this, everyday bottlenecks disappear. Take sprint planning and status updates, for example. Normally, a manager asks a team lead for an update, the lead asks the team, and everyone tries to remember what they did over the last two weeks—like a slow game of telephone.
In a queryable company, an AI agent simply looks at the raw data: the project tickets, the code commits, the customer support logs. It doesn't guess or rely on vague summaries; it tells you exactly what shipped and how it's performing, helping the team plan their next steps with much higher accuracy.
A New, Flatter Team Structure
When information flows smoothly without needing layers of people just to pass messages back and forth, our org charts can change dramatically. Instead of a deep hierarchy, we can organize our teams around three practical roles:
- The Builder (Individual Contributor): These are the hands-on executors across the company—whether in engineering, sales, or support. Because the AI system gives them all the context they need, they can make decisions and build things without waiting for a manager to tell them what to do.
- The Strategist (Directly Responsible Individual): Instead of managing people's daily schedules, this person owns a specific outcome or customer problem. They focus entirely on the "what" and the "why," bringing together resources to solve cross-cutting issues.
- The Player-Coach (Active Leader): This role replaces the traditional manager. Instead of sitting in alignment meetings all day, player-coaches actively build alongside the team while focusing on developing their people. (If you are a founder, this is you—you have to use these tools yourself; you can't just outsource your AI strategy).
Keeping Things Efficient: Token ROI
As companies adopt this, there's a temptation to just throw massive amounts of data at an AI model and rack up a huge API bill, thinking that means you're being innovative. But treating raw token usage as a badge of honor is a mistake—it just replaces human waste with digital noise.
The goal should be efficiency:
- Targeted context over data dumps: Don't feed an entire codebase or company history into a prompt. Give the AI agent only the specific files or data points it needs for the task at hand to cut down on costs and mistakes.
- Smart routing: Send simple, high-volume tasks to cheaper, specialized models, and save the expensive, advanced models for complex decisions.
- Quality over quantity: A few clear, well-thought-out instructions will always get better results than millions of generic words generated by a model.
Transitioning to this way of working will take some trial and error, and things will likely break along the way. But the best way to understand this shift is to jump in. Spend some time working directly with these tools and see how they can help your team communicate, coordinate, and build better together.
Sources & Further Reading
https://www.youtube.com/watch?v=EN7frwQIbKc
https://sequoiacap.com/article/from-hierarchy-to-intelligence/