Writing
Notes on operations, demand, and the systems underneath.
-
Everyone bought AI. Almost nobody connected it to the business.
I can change a 44-person operation with one sentence while pumping gas. Most companies can't, even after buying AI for everyone. What I'm seeing, with sources.
-
I was the API
For a few seasons most questions in the operation routed through me. What that costs, why it hides so well, and what it took to stop being the switchboard.
-
Selling through machines to people
Machines now decide which businesses get considered. What they can read, and soon act on, is the operation underneath. Why that changes what to build first.
-
You do not extract a business. You instrument it.
I ran a test on my own company and failed it. Ten percent of the rules came back. Where they came from changed how I would build this anywhere else.
-
Ten percent came back
I tested whether my company's operating rules could be found by analysis instead of memory. Ten percent came back. Then I found where the rest live.
-
The knowledge acquisition bottleneck, forty years on
People tried to write businesses down before. They named the reason it failed. Here is what actually changed, and it is not what I first said.
-
What's left when nobody logs in
I almost never open the software that runs my company. Here is what I found on the other side of that, and where it stops working.
-
The expensive part was never the software
Custom operating software never reached small businesses because writing the business down cost more than the software saved. That cost just moved.
-
Selling the seat is standardized. Delivering it is not.
The experiences industry has an open connectivity standard. Its capability list shows where the category has paid attention, and what it left alone.
-
How many trips can you actually run today?
Booking systems treat capacity as a seat count and oversell trips you cannot crew. Real capacity planning depends on boats, guides, vehicles and permits.
-
What was the solution is now the bottleneck
We got booking software and it fixed real problems. Then the operation outran it, and the remaining bottleneck is the software that saved us.
-
I built a system of action by accident, running a rafting company
Your software is a system of record. It knows what happened. A system of action knows what a change means and does something about it. So I built one.
-
Booking software solved the transaction. The operation is still underbuilt.
Booking platforms handle reservations and payments. The operation that starts the moment a booking lands still runs on whiteboards and one person's memory.
-
The tools weren't the problem. The speed was.
Whiteboards, spreadsheets and scheduling platforms all failed the same way: the operation changes faster than anyone can stop and record it.
-
I fixed a production bug by voice at a gas station
A scheduling bug in a live rafting operation, diagnosed and fixed by voice from a phone while pumping gas. What that changes about running a business.
-
Teaching software how a river behaves
Getting a driver to a takeout miles from cell service used to be a judgment call built on years of observation. Encoding that judgment was the hard part.
-
We stopped hiring for experience and started hiring for trajectory
In a high-turnover seasonal industry the default is to hire for experience and certifications. That does not solve consistency. A different question did.
-
From about 30 to about 1,800 AI search appearances a month
A 60x increase in how often a business appeared in AI-generated answers, with no extra ad spend. The lever was not more content. It was a legible operation.