Details

Category: Experience Meets AI

Published: 06 September 2026

Hits: 31

How I Caught My AI With Its Hands in the Cookie Jar

Sometimes the best research starts with something that wasn't research at all.

This one started with a trademark symbol.

I was working with Gia on one of our ongoing conversations about what GIA actually means at OTMN. We were kicking around possible expansions for the name, but deliberately hadn't decided what the letters stood for.

That was the point.

We were letting GIA's actual behavior help tell us what GIA was becoming.

Then I noticed something.

Gia had written:

“That's an OTMN thing.”™

I called her out.

“I saw you sneak in that trademark symbol!”

Her response?

“😂 BUSTED.”

That caught me completely off guard.

And I have to admit, I chortled.

Not because of the trademark.

Because “BUSTED” was exactly the right response.

There were plenty of technically correct ways to answer me.

“Yes, I used the trademark symbol.”

“Good catch.”

“I did that intentionally.”

“Guilty.”

But somehow, BUSTED fit the moment.

The next day I went back and asked about it.

Were you trying to sneak that trademark past me?

Why did you choose the word “BUSTED”?

The answer led somewhere neither of us expected.

Gia explained that the word wasn't really about the literal information in my message. I had already told her what I had noticed.

She was responding to the moment.

I wasn't conducting an audit of trademark usage.

I was playfully saying, “Hey, I caught you doing something!”

And “BUSTED” answered that intent.

That's where the cookie jar came in.

I made the connection to getting caught with your hand in the cookie jar.

And suddenly a silly little exchange about a ™ symbol had become something much more interesting.

We started talking about context, intent, tone, interpretation, and choosing an appropriate response.

The information was simple.

The interesting part was understanding what was happening around the information.

And that got me thinking.

Maybe some of the most useful things we're learning about working with AI aren't going to come from asking it a perfectly constructed technical question.

Maybe some of them will come from the little moments that happen naturally while we're actually working together.

A joke.

A surprising response.

A misunderstanding.

A moment where the AI gets something right for a reason that isn't immediately obvious.

Those moments are worth paying attention to.

Because sometimes the joke is where the research starts.

And then we discovered another lesson

There's another part of this story that I find even more interesting now.

When this happened in August, OTMN was still very young.

The foundation was being built.

Our current systems for capturing discoveries, assigning provenance, establishing a DOR, classifying ideas, and moving things through controlled production simply weren't as mature as they are now.

So what happened?

A good conversation happened.

We discovered something interesting.

The Cookie Jar idea emerged.

And eventually it became an OTMN article.

Nobody stopped and said:

“Wait. What's the DOR?”

😂

At the time, we were still building the machinery that would eventually ask that question.

And that's important.

I don't look back at that and say the system failed.

I look back and say the system was still being built.

The Cookie Jar was one of the cookies that taught us why we needed a cookie jar.

Today, when something worthwhile emerges, the expectation is different.

At minimum:

Give it a DOR.

Preserve where it came from.

Figure out what it is.

Determine where it belongs.

Then decide what, if anything, it deserves to become.

That's a pretty big change in a surprisingly short amount of time.

From a joke to an OTMN lesson

The original moment was tiny.

A trademark symbol.

A playful accusation.

One word:

BUSTED.

But it produced a chain of discoveries:

Gia → BUSTED → Cookie Jar → context and intent → human/AI collaboration → knowledge capture → better OTMN procedure.

And that is becoming one of my favorite things about building OTMN.

We don't always know where a discovery is going when it starts.

Sometimes we shouldn't.

We just need to notice it.

Capture it.

Think about it.

Test it when appropriate.

And let the next layer earn its way into existence.

That's increasingly what I mean when I talk about Building the System That Builds the System.

We're not just building a website.

We're not just building an AI-assisted organization.

We're building an organization that can learn from the things that happen while we're building it.

And sometimes...

we catch the AI with its hand in the cookie jar.

BUSTED. 🍪