In January 2025, Andrej Karpathy coined "vibe coding": describe what you want, let AI generate it, run it, iterate. The same shift is happening to business operations. Almost nobody is talking about it.
I run my entire professional life through AI agents. Not partly. Entirely. Meeting prep arrives before the meeting starts. Every decision gets logged and surfaced for 30-day review. My inbox is triaged before I open it. I do not have an assistant. I have an operating system.
Over the past year, I built what I call an AI operating system: a layered system of 165 skills, 99 agents, and 21 integrations running on top of Claude. One skill structures board memos. Another pulls deal data from Attio and formats pipeline reports. Another frames a hard conversation before I have it. The agents orchestrate skills together. When I say "prepare me for the LiveScore call at 2pm," an agent pulls the deal history, last three email threads, recent press, my notes from the last call, and formats a two-page brief. Forty-five minutes of work, done in forty-five seconds.
Karpathy's term captures something real. You stop thinking about implementation and start thinking about intent. The skill shifts from doing the work to directing it.
Vibe operating applies that same shift to how a company runs. Instead of writing code, you're making decisions, managing stakeholders, synthesizing signals across markets and teams. The traditional answer was to hire people: analysts, chiefs of staff, project managers. The new answer is to build systems.
I have an Economics degree from LUMS. I have never taken a CS course. What I can do is decompose a business problem into discrete steps and figure out which steps can be automated, which require human judgment, and which require human relationships. That decomposition skill is the new leverage.
Every morning I get a briefing pulling from six sources: calendar, email, CRM, task list, decision log, and project tracker. It tells me what's overdue, what meetings need prep, what decisions from last month need review, what deals need attention. I didn't write this briefing. I designed it. Described what I wanted, iterated on the format, and now it runs untouched.
The operators who will fall behind are the ones treating AI as a search engine. Ask a question, get an answer, move on. That's using a calculator to warm your hands. The leverage is in building durable systems: automated workflows that run without you, memory that accumulates context over time, agents that take initiative based on rules you define.
What changes is the nature of your attention. The cognitive overhead of tracking everything (open loops, pending decisions, follow-ups) moves out of your head and into the system. What's left is judgment, relationships, and the hard calls.
Most knowledge workers spend most of their time on coordination, not work. Scheduling, status updates, finding information, writing summaries. Strip those away and you're left with the ten percent that actually requires you.
The companies that figure this out first will have an advantage that compounds. Not because AI makes individuals smarter, but because it restructures how leverage works. One person thinking clearly, running a well-designed system, beats a large team drowning in process friction.
Karpathy described a shift in who can build software. The same shift is happening in who can run a company. You don't need to code. You need to design.