Price's Law, AI agents, and why your A-players are the leverage point
Price's Law says the square root of your team produces half the output. AI agents don't replace your A-players. They multiply them. Here's how that plays out inside middle-market DFW operations.
Every operator I've worked with eventually notices the same uncomfortable pattern. A small handful of people on the team are responsible for most of what actually moves. It isn't favoritism, and it isn't a morale problem. It's a math problem with a name. It's called Price's Law, and it has a lot to do with where AI actually pays off inside a middle-market business.
Price's Law in plain English
Derek de Solla Price observed that in any productive group, half of the output comes from roughly the square root of the number of contributors. Ten people on a team? About three of them are doing half the meaningful work. Twenty five people? Five. A hundred? Ten. The curve gets steeper, not flatter, as you grow.
You don't need to take the math literally to feel it. Walk any office, shop floor, or job site and you can name the people whose absence would actually hurt this week. That short list is your A team. Everyone else is necessary, but most of the leverage lives there.
The wrong instinct: hire more average
When output stalls, the default move is to add headcount. It rarely works the way leadership hopes. The bottleneck isn't usually hands. It's judgment, context, and how many decisions a person can make in a week. Adding average performers spreads the same A players thinly across more coordination, onboarding, and review. Output creeps up while payroll jumps.
The honest question isn't, "how do I get more people?" It's, "how do I get more out of the three to ten people who already carry this place?"
Where A-players actually lose hours
When we audit workflows, A players almost never lose time to the work that earned them their reputation. They lose it to the friction around that work:
- Re explaining the same context in email, Slack, and meetings.
- Hunting for a number that lives in three systems and one PDF.
- Producing status updates, summaries, and handoffs for everyone downstream.
- Drafting the first version of a quote, scope, response, or memo from a blank page.
- Babysitting workflows that depend on someone else doing their part on time.
None of that is the work you hired them for. All of it is exactly what well scoped AI agents are good at.
What an AI agent actually gives an A-player
Here's the mental model that works. An agent isn't a replacement for a person, and it isn't a chatbot bolted onto the side of your business. It's a piece of software that does the unglamorous 70 percent of a workflow so a human only has to apply judgment to the 30 percent that matters.
For an A player, that shows up in four places:
- Drafts instead of blank pages. Quotes, scopes, client replies, and weekly reports all start at 80 percent done.
- Retrieval instead of digging. "What did we quote this customer last year, and what did they actually order?" Answered in seconds, not twenty minutes.
- Handoffs that don't need nagging. The agent notices when a step is overdue and either chases it or escalates cleanly.
- A working memory. Notes, decisions, and context that used to live in one person's head become a system anyone can query.
Stack those up and the math gets interesting. If an A player goes from twelve good decisions a week to twenty five, you didn't gain a person. You gained most of one. That's the kind of result Price's Law predicts, and it's the kind hiring rarely delivers.
A concrete DFW example
One operations lead at a Dallas industrial supplier was the bottleneck on every outbound quote. Customers waited a day or two because she was the only person who could reconcile pricing tiers, stock, and freight against the customer's history. A scoped agent now drafts the quote from the inbound RFQ, attaches the relevant history, and flags exceptions. She approves or edits in minutes instead of rebuilding from scratch. Same A-player, more decisions per week, no new hire.
That's the pattern worth chasing inside a middle-market business. Don't aim AI at your average performer. Aim it at the three to ten people who already carry the place, and watch what happens to the curve.