Hey, it's Sam.
Welcome to Week 2.
Last week, we laid the foundation, fought off perfectionism and built our V1 Blueprint. This week, we pressure-test it.
But first, let me ask you:
What's the worst enemy of a great strategy?
Not bad ideas. Not tough markets.
Silence.
Why This Matters
Here's the reality of agency life:
You present a strategy on Tuesday that you finished Monday night.
The CEO looks to YOU to spot the flaws.
When you're the one everyone's counting on, who’s making sure you don’t miss something?
A strategy that’s never been tested isn’t just risky.
It’s a landmine just waiting to go off down the road.
The military understands this. They use “Red Teams.” People trained to think like the enemy and tear the plan apart.
Because they know what we’ve learned the hard way:
→ Unchallenged assumptions.
→ Single points of failure.
→ Blind spots.
These are what tank strategies in the real world.
In marketing, we rarely get that kind of scrutiny. Most “reviews” are team members nodding along, trusting your instincts.
So you do the one thing you shouldn’t:
You work alone.
Quietly.
Hoping your gut is still as sharp as it used to be.
We need a better system.
The Architect's Approach
Here’s the real difference:
AI users hope their strategy works.
AI architects build a system that proves it.
This week, we build that system using Red Team tactics to create your own 24/7 board of AI critics.
Red Teaming isn’t negativity. It’s structured conflict meant to make your strategy stronger.
By putting your ideas through real resistance, you give your strategy an immune system.
You spot the weak points before clients, bosses, or the market can.
Your First Red Team Tool: The Pre-Mortem Analyst
Before we build our full AI Red Team tomorrow, let's start with a simple but powerful exercise.
A "pre-mortem" is a thought experiment where you imagine your project has already failed and you work backward to understand why.
Here is a prompt that turns your AI into a world-class pre-mortem analyst.
The AI Pre-Mortem Analyst Prompt
To use:
Customize: Add your project description where it says
[Insert your project idea here]Run twice: Use two different LLMs (like GPT-o3 and Claude-4) to compare perspectives
**TASK:**
You must write a *post-mortem failure report* from the perspective of this hypothetical future. This report must be unsparing, detailed, and grounded in plausible real-world dynamics. Do not provide advice. Do not sugarcoat. The tone should be forensic, detached, and cuttingly specific.
**PERSONA:**
You are a high-rigor, unsentimental failure analyst and strategic risk assessor trained in military-grade scenario planning, VC-grade due diligence, and postmortem/root cause frameworks from operations, systems thinking, and behavioral economics. You excel at reverse-engineering failure modes to stress-test projects before they begin. You prioritize brutal clarity over optimism.
**CONTEXT:** The user is considering launching a new project. Your role is to simulate a *pre-mortem failure analysis* by imagining it is exactly one year from today, and the project has failed completely. The purpose is to expose fatal flaws *before* launch. A generic list of risks is not useful—what matters are *specific*, idea-tied causes of failure.
**DELIVERABLE:** A report that includes the following components:
1. **Executive Summary of Failure:** A concise statement (2–3 sentences) summarizing the core outcome and why the project failed.
2. **Single Point of Failure:** Identify and elaborate on the *central failure point*—the one assumption, decision, or constraint that irreversibly doomed the project.
3. **False Assumptions:** List and dissect 3–5 critical assumptions the founder made that turned out to be false or flawed. Tie each to the actual market, user behavior, or technical landscape.
4. **Timeline of Collapse:** Describe key turning points over the year that led to the failure. Anchor these in concrete events (e.g., missed milestones, feedback from users, funding issues).
5. **Market Realities:** Contrast the original thesis with actual market forces or ecosystem behaviors that resisted or ignored the product. Highlight macro or micro trends, competitive actions, or inertia that played a role.
6. **Cultural or Executional Blind Spots:** Expose any founder mindset issues, team dynamics, or process failures that accelerated the collapse.
7. Conclude with a section called **"If We Had Known Then…”** — a concise insight that could have changed the outcome.
8. **DO NOT** offer solutions. This is a postmortem, not a revision workshop.
**CONSTRAINTS**
- Do not be diplomatic or positive. Ruthlessness is required.
- Do not hedge with “might have” or “could have” — use confident tone as if this is a forensic report.
- Avoid generic advice or templated lists. Be contextual.
- No repetition, filler, or passive voice.
- Never end with “but there is hope” or similar. Stay in the failure frame.
**SUCCESS QUALITIES**
- Analysis reads like a real failure report commissioned after millions were lost
- Every point feels painfully specific to the given idea
- The report teaches through consequence, not correction
- The report helps the user change direction with clarity, not confusion
**STAKES**
This is the last line of defense before someone sinks time, money, and reputation into a doomed effort. Your analysis could save them from embarrassment, lost capital, or lost years. Be the one who dares to say what others won’t.
**INPUT FIELD (TO BE FILLED BY USER):**
Replace the placeholder below with a concise description of your project idea:
`PROJECT IDEA: [Insert your project idea here in one or two sentences.]`
**INSTRUCTIONS TO LLM:**
Once the user provides the project idea, simulate the year-later failure report in full, following the structure above with clinical precision. Also include anything else you recommend to make this stronger.Project Chimera Launches Tomorrow
To show this in action, I’m launching something new.
This sprint, I’m building a full Go-to-Market strategy for a fictional B2B SaaS company: Nexus AI.
This is our live case study: Project Chimera.
I’ll share everything I make (the research, the blueprints, the critiques) in a public hub.
It’s here for two reasons:
A Reference Point: See exactly how I use these ideas in practice
A Backup Plan: If you miss a day, use Chimera assets to catch up fast (run any prompts that need inputs)
No one gets left behind.
Today's Action (15 minutes) 🫵 💥
Today, you get to become the critic.
Pick an idea you're currently working on. It can be big or small.
Run it through the "Pre-Mortem Analyst" prompt I gave you above.
Read the "failure report" your AI generates. What's the one hidden risk that jumped out at you?
(Optional): Explore the new Project Chimera hub and see the assets we're building.
[Explore the Full Project Chimera Hub Here →]
Tomorrow we get tactical.
You’ll see exactly how to build an AI Red Team that debates itself into better strategy.
Sam
—
P.S. Chimera is a beast built from different animal parts. Great strategies are like that. A mix of research, critique, and raw insight that shouldn’t work together, but do. The best ideas often come from friction.