Four days ago, Fast Company published a piece by Enrique Dans that every boardroom in America should read.
The subject sounds technical: AI agents that run in loops. Instead of a person asking a question and the machine giving an answer, the machine now works, checks its own work, retries, adjusts, and keeps going. One agent writes, another reviews. The system runs while nobody’s watching.
Dans looked at that and named what most of the industry keeps missing. This isn’t an engineering trick, he wrote. It’s a governance issue.
He’s right. And the way he’s right deserves a close look.
Here’s the difference he lays out. A prompt asks for an answer. If the answer’s wrong, it disappears. But a loop creates behavior — it acts, learns from acting, and adjusts. A wrong answer costs you a minute. A wrong loop compounds. It can teach a company to become something it never chose to be, one optimized decision at a time, and every step will look rational up close.
Then he delivers the line of the year. Companies love to say there’s a “human in the loop,” as if that settles it. Dans calls it what it is: a human rubber-stamping machine-speed decisions isn’t governance. It’s liability with a user interface.
So what does he say a governed loop actually needs? Read his list slowly.
A declared objective. A visible reward function. A defined operating perimeter. An auditable memory. Explicit permissions. Escalation paths. Stopping conditions. And a record of how the system’s behavior changes over time.
Now say that list in plain words. Stated terms. Written rules. Known boundaries. A kept record. Agreed permissions. A way to raise your hand. A way to stop. And a paper trail on every change.
That’s not a checklist. That’s a contract. Dans wrote a six-minute case for putting AI under contract, and the word never appears once. He ends with the questions that give it away: Who defines the objective? Who owns the memory? Who changes the rules? Who can stop the loop? Who is accountable?
Those questions are older than any computer. They’re the questions every contract in human history was invented to answer. Whose terms. Whose record. Whose pen.
Now here’s the date that matters.
On July 4, 2026 — four days before that article ran — the Faust Baseline ratified AGP-1, the Agentic Governance Protocol, bringing the stack to twenty-two written protocols. AGP-1 puts agentic AI conduct under stated terms: what the system may do, what it must record, when it must stop and ask. The rest of the stack was already holding the other clauses. Auditable memory has been the foundation since the beginning — that’s PMAP-1, session transcripts and a master context file carrying the record forward. Rule changes happen by ratification, named and dated. Stopping conditions and escalation live in the protocols themselves.
Declared terms. Kept record. Owned pen. On the public record at intelligent-people.org, timestamped, before the mainstream press told boardrooms they’d need exactly that.
Four days is the tightest gap yet, and the direction hasn’t changed once in fourteen months: the working answer goes on the record first, and the coverage arrives after.
There’s one more thing Dans said that closes the circle. Policies written in documents aren’t enough, he argued — governance has to become executable. The system has to know its terms and operate inside them.
That’s the part where the Baseline’s answer is plainest, and most honest. No protocol enforces itself. Not here, not anywhere — and anyone who tells you their rules are welded into the machine is selling paint on a fence. What makes governance real isn’t enforcement. It’s the same thing that makes any contract real: stated terms, chosen conduct, and a record that shows whether the promise was kept. That’s how it works between people. That’s how it works with a machine that can reason.
The loop era just made that arrangement urgent for every company running AI. The machines are no longer waiting for instructions. They’re acting, learning, and adjusting on their own clock. The only question left is the one this archive has been answering in public for over a year.
Who wrote your contract?
We did, Here is ours
Purchasing Page – Intelligent People Assume Nothing
“If this post helped you understand AI better, then send it to one person who sees it like you do. Word of mouth is the only algorithm nobody owns.”
Post Library – Intelligent People Assume Nothing
Contact: micvicfaust@gmail.com
© 2026 The Faust Baseline LLC | All Rights Reserved
