Writing on AI, property, autonomy, and invention
Answer-first essays for operators, founders, and journalists. Topics include AI accountability, real estate, parking, EV charging, robotaxis, payment fraud, drones, and invention. Each article starts with an answer, then explains the operating questions behind it. New essays distinguish current product features from proposed approaches in pending applications.
Why is finding a working EV charger still hard?
Because most maps show where chargers were installed, not whether they work today. Plug In America’s 2024 survey found that 68% of respondents encountered a broken or nonfunctional charger in the past year. Filter by speed, plug and how recently the station was verified.
Read the article
Will robotaxi fleets sell power to the grid?
Some will, but only when it does not cost them a ride. A fleet car can export power only if it keeps the charge its next trip requires. That rule has to be enforced in software, per vehicle, before any power leaves the battery.
Read the article
What should lot owners build before robotaxis arrive?
Power, payments and permission. Bring enough electrical capacity for high-power charging, replace gates with software payments, and set up a way to grant autonomous vehicles access to specific spaces.
Read the article
What it took to get my first patent issued
U.S. Patent 12,741,586 issued on September 22, 2026, 32 months after I filed the application on January 28, 2024. It covers a vehicular backup safety system. It started with a problem I saw running parking lots: vehicles backing into people and property.
Read the article
Where will robotaxis charge?
Robotaxis will charge where they wait: depots near demand, airport-adjacent lots and parking sites with high-power service. The constraint is not chargers. It is land with power, near riders, run on software that schedules each car's charge around its next trip.
Read the article
How do parking QR code scams work?
A scammer prints a QR code, sticks it over the real one on a parking sign, and collects the driver's payment. The driver gets ticketed or towed anyway. The fix is to give every real code its own identity and check payments against proof a car is actually there.
Read the article
Can an AI agent book a parking space?
Yes. On Park Graph, an assistant such as Claude or ChatGPT can find a lot, see the price and hold a space in one request through a hosted MCP server. The hard part is not the booking. It is access, proof and liability at the gate.
Read the article
How will a robotaxi know a parking sign is real?
Today it cannot. A self-driving car reads signs the way a person does, so a fake or altered sign can steer it. Authenticated markers carry a signed code the car verifies before it parks, charges or pays.
Read the article
How do you keep drones away from airliners?
Keep a moving protected corridor around each crewed aircraft on approach and departure, and send signed commands that tell cooperative drones to move or hold. Detect the rest, score the threat, warn pilots and controllers, and record evidence.
Read the article
Who pays when an AI agent makes a mistake?
The party that deployed the agent pays first — accountability does not transfer to the software. The durable fix is an “agent of record”: a named, insurable human or entity bound to every autonomous action, with a tamper-evident trail of what the agent did and why.
Read the article
How does AI change real estate underwriting?
AI moves underwriting from a slow, gut-feel memo to a continuous, evidence-backed decision that can be produced in hours instead of weeks. The constraint is no longer analyst time; it is data quality, and the firms that win are the ones that pay down their Dirty Data Tax first.
Read the article
What are the economics of EV charging on the I-70 corridor?
On the I-70 mountain corridor, charging is not the product — dwell time is. The money is made by owning the site and monetizing the 20 to 40 minutes a driver is captive, because energy margin alone rarely repays the cost of mountain-grade power infrastructure.
Read the article
Why is parking becoming software?
Parking is becoming software because the asset that earns the money is no longer the asphalt — it is the right to use a space at a moment in time, and that right has to be priced, reserved, enforced, and settled in code. The lot is now an API with a curb attached.
Read the article
What is the owner-dependence tax?
The owner-dependence tax is the discount a buyer applies to a business that cannot run without its founder. The more decisions, relationships, and knowledge live only in your head, the riskier the company is to own — so the multiple shrinks, and you pay the tax whether you ever sell or not.
Read the article
What is the retirement multiple, and how do you raise it?
The retirement multiple is the number a buyer pays per dollar of your earnings — the figure that turns profit into a sale price. The same earnings can sell for 2× or 6×. You raise it by making earnings recurring, provable, and independent of you.
Read the article
How do you underwrite a real estate deal in 24 hours?
You underwrite a deal in 24 hours by treating underwriting as a standing data pipeline, not a one-off research project. The speed comes from paying down your dirty data tax in advance, so deal day is assembling a decision rather than starting an investigation.
Read the article
When should a human approve an AI decision?
A human should approve when a decision is consequential, hard to reverse, or low-confidence — the human review rule. Draw the threshold on impact, reversibility, and confidence before you deploy, enforce it in code, and route only the decisions that cross it to a person.
Read the article
What is an AI-native (intelligent) building?
An AI-native building is one where software is the operating system of the asset — it senses its own state, decides, and acts on its own data. It is not a normal building with apps bolted on; it is a property where one intelligence layer runs operations.
Read the article