Welcome back to the Launch Key 🚀

Last week George Pillari asked whether you could still fly the plane if AI handed you the stick. Good question. Here's the follow-up: what happens when the flight rules change while you're in the air?

I got a phone call in the mountains last week that sent me digging through my own digital files. I found something. I think you've got stuff too.

Let's get into it.

An old friend called last week wanting blockchain advice.

I had to tell him I've moved on. Which is true. I spent six years on it, built a company, took the swing, and I'm not going back.

Then I hung up and spent the rest of the morning realizing I'd told him something that wasn't quite right.

Here's what we actually built.

In 2021 we filed a patent for something we called a contract codex. The idea: instead of writing rules into a contract and hoping they stay true, you keep the rules somewhere else — in a queryable database of federal, state, local, and industry requirements — and the contract checks that database on a schedule. When a regulation changes, the contract notices, tells the parties, and generates an updated version of itself that complies. It issued as US 11,557,008 in January 2023.

Here's what I said about it at the time: "This patent adds contingency for regulatory or other conditions that may change over time."

It was built for natural gas transactions. A staggeringly boring problem in a staggeringly regulated industry. The market never came, for reasons I'll write about another time.

Look at what’s happening right now.

Three weeks ago, on August 2, the EU AI Act's transparency and watermarking obligations went live. Meanwhile, the deadlines for high-risk AI systems — the ones everyone spent two years building toward — got pushed again, from February 2025 out to December 2027, and to August 2028 for systems embedded in regulated products. The stated reason is almost funny: the Commission hadn't finished writing the standards companies were supposed to comply with.

On this side of the Atlantic it's worse, because it's fragmented. California, Texas, and Illinois all had AI laws take effect January 1. Colorado's landed in June. And a December executive order set up a federal litigation task force whose job is to challenge those very state laws.

So: the rules are moving, in opposite directions, on different clocks, in every jurisdiction you operate in. And your AI agent is out there right now sending emails, pricing deals, screening candidates, and making decisions on your behalf.

If you hard-coded your compliance rules into that agent, you are going to be rewriting it. Repeatedly. Forever.

The alternative is to separate the rules from the execution. Put the rules somewhere the agent can look them up. Make looking them up part of the job, not a quarterly project. Compliance as a feature, not a retrofit.

That's the codex. We built it for gas contracts in 2021 and couldn’t see all the possible future uses. It’s exactly what’s needed today.

This is the part I want you to hear.

I'm not telling you this because I was ahead of the curve on AI. I wasn't. In 2021 I was thinking about pipelines and NAESB contracts and a 3-month settlement process that still ran on Excel and emailed PDFs.

I'm telling you because I had spent thirty years watching companies bake assumptions into systems and then discover that the assumptions moved. That's not genius. That's just having been around long enough to have been burned by it. The pattern was the asset. The industry was an accident.

And here's an uncomfortable part: I couldn't see the pattern at the time. I could only see the gas contracts. It took a phone call from an old friend and five years of distance for the whole thing to come together for me.

You almost certainly have at least one of these.

Not a patent. But a proposal that got killed. A pitch your VP loved and your VP's boss didn't. A process you designed that got shelved when the reorg came. A product spec, a business plan, a fix you knew would work that never got funded — killed for reasons that had nothing to do with whether it was right.

You filed it away and moved on, because that's what professionals do.

The conditions that killed it are not the conditions we're living in now.

The Shelf Audit

Four questions. Twenty minutes. Do it this week.

1. What did you build, propose, or design that got killed for reasons that had nothing to do with the merits? Wrong timing, wrong boss, wrong budget cycle, wrong decade. Write down three of them. You'll be surprised how fast they come back.

2. Strip your industry out of it. Describe the mechanism in one sentence using no nouns from the business you were in. Mine went from "smart contracts for natural gas that update when FERC rules change" to "a system that keeps its rules somewhere it can re-read them." That sentence is the asset. The rest was context.

3. What's changed since? Technology got cheaper, a regulation appeared, a cost collapsed, a market grew up. Be specific. "AI is a big deal now" is not an answer. "The thing that made this cost $400K in 2019 is now a $20 API call" is.

4. Who has that problem today, and would they pay not to have it? If you can name three companies and one of them would take your call, you're not reminiscing. You're doing market research.

Most of what's on your shelf should stay there. That's fine — that's what an audit is for. But you have also been filing away good work for thirty years under the assumption that "no" meant "wrong."

Sometimes "no" just meant "not yet."

Go look.

Now go launch something 🚀

The biggest IPO in history is live — $1.75T valuation, $135 open, $75B raised. And history says the retail investors who chase day-one hype are the ones who get burned.

The SpaceX run created thousands of new millionaires. But it wasn't the people buying at the open. It was the ones positioned early, in the names set to ride the wave.

Our analyst pinpointed 3 stocks positioned to ride the SpaceX wave — with entry guidance and price targets, a bonus 4th pick (the most undervalued name in the sector), and a 3-phase playbook for when to buy and when to sell.

Over 2,500 investors have already read the report. Claim your free copy here.

Get the Free Report

The codex isn't a blockchain idea, and it isn't an AI idea. It's a separation of concerns — and David Parnas wrote the paper on it in 1972.

His argument, in On the Criteria To Be Used in Decomposing Systems into Modules, was that you should draw the lines in a system around the things most likely to change. Put the volatile stuff behind its own door so that when it moves, nothing else has to.

Fifty-four years old and it's the single most useful thing you can know about deploying AI agents in 2026. Regulation is the volatile part.

On the Criteria To Be Used in Decomposing Systems into Modules (PDF)

Gemini Notebook — yes, I’ve shared this before but it’s free, and built almost perfectly for your Shelf Audit.

Dump everything in: old decks, the business plan that didn't get funded, that 40-page strategy doc nobody read, your resume, whatever's in the folder. It reads across all of it at once and answers questions grounded only in what you gave it — no wandering off into general knowledge.

Then ask it the hard ones. "What problem do these documents keep circling back to?" "What did I propose more than once?" "Describe the mechanism in this deck without using any industry terms."

That last prompt is question two of the audit, and a machine with no memory of the politics is much better at it than you are. You remember why it got killed. It doesn't.

Here's an archive most people rarely think to open: patents are free, full-text, and searchable.

Google Patents and Justia hold millions of documents in which someone described, in careful detail, a problem they thought was worth protecting — and a lot of those problems are more relevant now than when they were filed. Expired and abandoned patents are public domain. Reading them is the cheapest competitive research there is, and it's a very good way to find out whether the thing on your shelf has already been solved, already been abandoned, or is still wide open.

Search your old industry. See what people were worried about in 2015. Ask what's changed since.

📕 MGMT Playbook : Practical management insights straight to your inbox every Wednesday.

🏢 Retirepreneur : Your guide to turning decades of experience into profitable consulting, coaching, or course businesses, without risking your retirement security.

🗃 Dealroom Business Success Uncovered : Learn directly from billionaire entrepreneurs on how to grow a business. Join a community of 2,000+ innovators.

🧾 Redefining Retirement : Too Young to Retire, Too Old To Take Orders? Yeah, Me Too.