When AI Makes Production Free, Substance Becomes the Only Moat
This week I ran an AI workshop in Munich and built a video studio on my laptop. Those two things sound unrelated. They are not.
Both taught me the same lesson: when AI collapses the cost of production, the advantage shifts entirely to the person with something real to say.
The Uncomfortable Shift Nobody Is Naming Clearly
For the past few years, the AI conversation in most boardrooms has been about capability. Can we use it? Is it accurate enough? What tools should we buy?
That question is now largely settled. Yes, you can use it. It is accurate enough for many high-value tasks. The tools are commoditized.
The harder question — the one that separates operators from spectators — is this: now that production is nearly free, what are you producing that is actually worth anything?
This applies to content. It applies equally to analysis, decision-making, and how you run a leadership team.
What a Room Full of PE Professionals Taught Me About Second Views
In Munich, I worked with a team that lives inside pressure, ambiguity, and incomplete information — that is their daily operating environment. Special situations, carve-outs, complex restructurings. They are not naive about risk.
What shifted the energy in the room was not a demo. It was the moment the team stopped watching AI and started building with it — taking their own real use cases and shaping them into working prototypes.
The exercise that stayed with me most was what I call the Devil’s Advocate Stack: instead of relying on a single AI analysis, we set up a second, third, and fourth independent agent to challenge the first output. Each one was tasked with finding weak assumptions, blind spots, and false confidence.
It was uncomfortable. In exactly the right way.
Because one model — like one advisor, one analyst, or one slide deck — can sound entirely convincing. A structured stack creates contradiction, cross-audit, and ultimately a better human decision at the end. That is not AI replacing judgment. That is AI sharpening it.
In PE terms: it is what a good investment committee is supposed to do before a thesis gets too polished to challenge.
The Video Studio Paradox
Separately this week, I built a local AI video studio on my own machines and canceled several subscription tools. The clip I produced — end-to-end, in under three minutes — would have required a crew and a budget not long ago.
The most interesting line in that video did not come from me. It came from the AI version of me:
“I can copy his face, his voice and his gestures. What I cannot copy is whether he means it.”
I did not write that line for effect. I found it unsettling when it came out. Because it is true — and it applies far beyond video.
When production becomes cheap, polish stops being a moat. In 2026, almost anyone can publish content that looks expensive, generate analysis that sounds thorough, or build a process that appears governed. The surface has never been easier to replicate.
What you cannot replicate is whether it is real. A lived point of view. A decision you made under genuine uncertainty. A principle you held when it cost you something.
The Framework: Substance, Structure, Sign-Off
Across both experiences this week — the workshop and the studio — I kept returning to the same three-layer test. I use it now before deploying AI on anything consequential:
The 3S Test for AI-Augmented Work
- Substance — Does the underlying input reflect real knowledge, real context, real stakes? AI amplifies what you bring. If you bring noise, it returns polished noise.
- Structure — Is there a second view built into the process? A Devil’s Advocate layer, a cross-audit, a deliberate contradiction before you commit? One clean output is a liability, not an asset.
- Sign-Off — Who is accountable? AI can generate the analysis. A human — with skin in the game — has to own the decision. That line cannot be blurred.
This is not a framework for AI specialists. It is a governance habit for any operator who wants to use intelligent systems without outsourcing their judgment to them.
A Note on Servant Leadership and Why It Connects
This week I also received a recognition from my YPO network — something I am genuinely grateful for, though not because of the badge itself.
What it reminded me is that servant leadership and AI adoption share the same foundation: you have to earn the room before you can change it.
Empowerment beats done-for-you. In the Munich workshop, adoption accelerated the moment smart people built their own tools and saw that AI was not a productivity layer someone else operated — it was a new way to structure their own thinking.
Leaders who adopt AI the same way they would buy a software license — passively, at arm’s length — will get productivity theater. Leaders who build understanding first, who sit in the discomfort of the Devil’s Advocate setup, who ask “what am I actually trying to decide here?” — they will build a genuine edge.
The room does not get stronger because you bring AI into it. The room gets stronger because you bring better questions.
What You Can Do Next Week
-
Run a Devil’s Advocate exercise on one live decision. Take your current analysis or recommendation. Prompt a second AI agent with explicit instructions to find the three weakest assumptions. See what surfaces.
-
Audit one production process for substance. Whether it is a report, a presentation, or a content piece — ask honestly: does the underlying input reflect real knowledge, or are we polishing something shallow?
-
Name the human sign-off. For any AI-assisted output going to a client, investor, or board: make explicit who owns the conclusion. Write their name on it before it leaves the room.
-
Try the manual version first. Before automating a workflow, run it by hand once. Understand the friction. Then decide what AI is actually solving. This is the discipline I call treat AI like cash — don’t deploy it until you know what it is replacing and why.
-
Ask your team one direct question in your next meeting: “Where are we relying on one view when we should be building in a second?” You don’t need an AI agenda to start that conversation. You need honesty.
When production costs collapse — in content, in analysis, in operations — the only durable moat is whether what you are building is actually grounded in something real.
That has always been true. AI is just making it impossible to ignore.
What I’m curious about: In your business, where do you currently rely on a single view — one analyst, one advisor, one model — before a significant decision? And what would it take to build a structured second view into that process?
I’d like to hear where you’re starting.