Today's entry has a twist. The agent flagged something genuinely useful, and the action it recommends raises a strategic question I don't yet have a clean answer to.

So I'm bringing the question to you.

What Instant Crush told us this morning

Score: 6 out of 10. Steady, not weak.

The agent's full read:

"This area scores 6 out of 10 and is one of the steadier parts of the picture. The product language points to daily insight and monthly reframing, and the Brand Log shows recent activity in May 2026. The pattern here is positive: the business appears to be iterating in real time. That matters because buyers want decision support that stays current. But there is a difference between visible activity and visible method. Without a clear review framework, the brand can still look personal and opinion-led rather than system-led."

And the summary table:

Position now: You look active and current, but your refresh method is still mostly implied.

Positive signal: The product and Brand Log both suggest a real habit of ongoing review.

Attention point: Buyers can see activity, but not the system behind how you update decisions.

Next step: Make your review cadence visible with a simple market and competitor check format.

Competitive effect: This helps you look alive and current, but stronger systems build more trust.

That last line — stronger systems build more trust — is the one I want to sit with.

The distinction the agent is making

The distinction the agent is making

Activity is visible. You can see we post on LinkedIn, we update the website, we ship product changes, we publish this blog. The Brand Log itself is evidence that something is happening.

Method is invisible. Buyers can see that we update. They can't see why we update, what triggers a recheck, or how we know a shift is needed versus a tweak.

Without method, even good iteration can read as taste. As opinion. As "Rolf had a feeling on Tuesday." Which is a vulnerable position, because taste is replaceable. Method isn't.

This matters more for Instant Crush than for most brands, because our whole positioning is better judgement — which means we have to look more disciplined than competitors, not just busier.

The agent's point: we're iterating, but we look like we're iterating on instinct.

We're not.

What's actually under the hood

Here's the part I haven't shared publicly yet.

Instant Crush runs on what we call the Core. It's the context layer the agent uses to make recommendations. It's not just system prompts — it's a structured set of principles, evidence, decision frameworks, and meta-rules that have been refined over months of building and testing.

It has levels. It has a logic. It has a hierarchy of what overrides what when things conflict. It's the reason the agent doesn't just hallucinate marketing advice — it operates inside a structure we've spent a long time designing.

That structure is what makes the difference between "AI that writes marketing-flavoured text" and "AI that does diagnosis, recommendation, and briefing in a methodical way."

Three big strategic shifts in our business have come directly out of working through the Core. Dozens of smaller ones have too. When the Core changes, the recommendations change. When market signals shift, we update the Core, and the change propagates through every diagnosis the agent does afterward.

The system exists. We just haven't shown it.

The question I'm stuck on

The agent's recommendation: make the cadence visible.

That's the easy interpretation. Show what we review monthly. Show what triggers a recheck. Show how competitor and market signals flow into our thinking. This is doable. I can write that up this week.

But the deeper question is: how much of the Core do we make public?

There's a real argument for opening it up:

• It's the strongest possible proof of method. Right now buyers see "smart opinions." Showing the Core would shift the brand from opinion-led to system-led in a single move.
• It compounds trust. Method is replicable in principle, but very hard to replicate in practice. People are usually impressed by how much behind something there is — not threatened by it.
• It positions Instant Crush as the standard for evidence-based marketing AI. Whoever publishes their method first owns the conversation.

There's an equally real argument for keeping it closed:

• It's our genuine edge. The surface of the product can be copied in a weekend. The Core can't — but the structure of the Core can be, and a structure is enough for competitors to start chasing.
• We've spent months refining it. Handing the headstart to fast followers feels strategically dumb.
• The big AI companies — including the ones whose models we use — would absorb the structure into their own offerings within months.

My current gut: show the cadence, not the contents.

That means making public what we review, when we review it, what triggers a rethink, and what kinds of signals feed into the Core  without publishing the actual prompts, hierarchies, or meta-rules.

Visible method. Hidden mechanism.

But I'm honestly not sure. There's a version of the build-in-public bet where the right answer is "show more, not less." And there's a version where the right answer is "the people who'd trust us based on the Core are also the people who would trust us based on the cadence  so don't give away the edge."

I keep flipping between the two.

What I'm doing this week

Two things, while I figure out the bigger question.

1. Publish the cadence.

A short, public document that describes:

• What we review weekly, monthly, quarterly
• What triggers an out-of-cycle recheck
• What types of market and competitor signals feed into the Core
• The three big shifts we've already made (with brief reasons)

This is the agent's actual recommendation. It's doable now. It doesn't risk anything strategic.

2. Run the Core question through diagnosis.

I'm going to ask Instant Crush itself: should the Core be public? Not because the agent will give me a clean answer, it won't, but because forcing the question through the same diagnostic structure we use for everything else will probably surface considerations I'm currently missing.

I'll share what comes out of that.