#10 When Regulars Open Restaurants

To open a restaurant, you used to need to be a chef. You needed to create recipes, source ingredients, run a kitchen, direct prep cooks, coordinate the team, manage costs. Loving food wasn't enough. You had to know how to make it.

To build an AI agent, you used to need to be a developer. You had to write code, connect APIs, design systems, debug. Understanding your domain wasn't enough. You had to know how to build.

Both barriers just fell. And the people walking through aren't chefs or developers — they're the ones who always had the skill that mattered most: knowing exactly what good looks like.

What Changed

This series traced it piece by piece. Kitchen equipment got cheap. Recipes can be built from accumulated knowledge rather than culinary school. Supply chains connect through a single standard interface. Prep cooks work when given clear instructions. Teams coordinate on their own. Quality checks run systematically. Costs can be optimized. You can pick the best expert in each area and assemble your own combination.

With all of that in place, the requirements changed — for restaurants and for agents alike. Before, you needed someone who could cook. Now, someone who knows and loves food is enough. MCP connects the supply chain. Skills are built by organizing domain knowledge. Subagents work when directed. The infrastructure that used to require a full engineering team is available out of the box.

What Regulars Bring

Think about the regular. The one who comes every week, studies the menu carefully, asks the chef questions, learns about wine, knows a fair amount about ingredients. Has a feel for what tastes good, what combinations work, what guests want.

If this person said "I want to open a restaurant" five years ago, the answer was: "Go to culinary school first." Knowing flavor and making flavor were separate skills. No matter how refined your palate, if you couldn't handle a knife, you couldn't open a restaurant.

Now that regular can open a restaurant. Not because they learned to cook — because the kitchen system handles the cooking. What they bring is what chefs often don't have. Years of experience dining at different restaurants — "this place did X well, that place fell short on Y" — experience accumulated as a consumer is domain knowledge. An instinct for what guests want, because they've been the guest. And the ability to say "exactly this" instead of "whatever" — which turned out to be the core of everything.

Their job isn't to cook. It's to decide what kind of restaurant to build, set the direction, and judge the quality of what comes out.

The Same Shift, Different Kitchen

Building an AI agent now requires the same thing opening a restaurant does: deep knowledge of your own domain. "Our team's reports need this structure." "Our customers want this." "The important part of this data is here." If you have that domain knowledge, you can build an agent. Not by writing code — by directing a system that handles the code for you.

Before, only chefs could open restaurants. Now, someone who loves food and knows it deeply can. The barrier was never taste — it was infrastructure. And that's the part that just got cheap.