type: content-brief status: seed source: Steve Cunningham webinar transcript (2026-04-03) + Kathryn's riff target: LinkedIn post (organic) angle: Constraint diagnosis vs. brute-force automation
Content Brief: Paving Every Road in Town
The Line
Automating everything is the AI equivalent of paving every road in town. Diagnosing the constraint is figuring out which road people actually need to drive on.
The Argument
The current AI conversation — especially in the consultant/agency world — has collapsed into a volume game. Build more agents. Build them faster. Build them overnight. The pitch is: give me your files, I'll give you 100 agents by morning.
That pitch works if you're selling agents.
It doesn't work if you're trying to fix a business.
Because speed without diagnosis just compresses the same broken system. You get the same bottleneck, running faster, producing the same mediocre output — now at scale.
The practice owner who automates everything still has the same constraint on Monday morning. They just have 100 new things to maintain on top of it.
The Tension
This is NOT an anti-AI argument. This is an anti-"automate everything" argument.
AI is the most powerful operational tool in a generation. But a tool without a target is just expensive motion.
The question isn't "what CAN we automate?" The question is "what's actually capping this business right now?" Build for that. One system. One constraint. Then measure. Then build the next one.
Who This Speaks To
- Practice owners who've been told they need agents but don't know where to start
- Practice owners who HAVE started automating and feel busier, not better
- Consultants who are pitching "we'll build your agents" without diagnosing anything first
- Anyone feeling the pressure to "do AI" without a clear reason
AOS Connection
This is the core of the Deploy Sprint and Advisory OS thesis:
- Diagnose the constraint first (Systems Diagnostic)
- Build the one system that breaks it (Deploy Sprint)
- Measure, then build the next one (Advisory OS)
The value isn't in the building. The building is trivial. The value is in knowing what to build.
Possible Hooks
- "Everyone's building agents. Almost nobody's asking what to build them for."
- "You can automate 100 things overnight. 99 of them won't matter."
- "AI doesn't fix broken systems. It scales them."
- "The most expensive AI project is the one that automates the wrong thing."
- "Paving every road in town doesn't help if people only drive on three of them."
Tone Notes
- Not anti-Cunningham, not anti-anyone. This is a positioning piece, not a takedown.
- Calm authority. Pattern-revealing. The reader should feel like they just realized something they already knew but hadn't articulated.
- No "crush it," no hype. Just the observation.
Next Step
Run through VIP /organic when ready to produce. Monday or later.