Hey, it’s Sam.
Yesterday I promised to show you how to build engines, not tasks. Time to deliver on that promise.
Let’s get right into it…
Here’s the uncomfortable truth: The most common way people use AI for research is also the fastest path to creating generic, forgettable strategies. It’s the “best practices” trap.
Why This Matters
Early in my AI journey, I was building a strategy for a new client. I did what most people do:
Asked ChatGPT for “best practices for B2B content marketing.” It delivered a pristine list:
“Create high-quality blog posts”
“Leverage email marketing”
“Use social media”
All correct.
All useless.
I felt that specific kind of dread that comes from realizing you’ve created something spectacularly mediocre.
Here I was, supposedly an “AI strategist,” about to present the same vanilla insights my client could have found in any HubSpot blog post from 2019.
My AI hadn’t accelerated my thinking.
It had just made me average, only faster.
That day taught me something crucial: The job of a strategist isn’t to recite best practices. It’s to find openings no one else sees.
The Architect’s Approach
Here’s the shift that changed everything: AI users ask for quick answers. AI architects demand deep research. And here’s what nobody’s talking about: Every major LLM now has deep research capabilities built in.
ChatGPT’s Deep research
Claude’s Research mode
Gemini’s deep dive (my personal favorite)
Even Copilot does Research mode now
You’re not married to one tool. You’re orchestrating an entire research team.
Stop prompting for “best practices.” Start triggering comprehensive research reports. Let me show you the difference:
The User’s Prompt (The Trap)
“What are the best practices for launching a new SaaS product?”
The Architect’s Prompt (The Engine)
“Conduct deep research on SaaS launches. Analyze the go-to-market strategies of 5 specific SaaS companies that reached $10M ARR within 2 years between 2020-2024. Include: documented tactics, founder interviews, launch sequences, pricing evolution, and early customer acquisition channels. Cite all sources. Create a comprehensive report with specific examples, not generalizations.”
Here’s what happens next: Your AI doesn’t just list tips.
It generates a 42-page research document.
Most people see that and think “ugh, too much.” I see it and think “perfect start.” Because that massive report?
That’s not your final strategy. That’s the seed of your strategy oak tree.
Your Deep Research Tool (The Architect's Prompt)
To get you started, here is the exact, copy-and-paste prompt template I use. This isn't a simplified version; it's the professional-grade tool. Replace the bracketed text with your own inputs.
=========================================================
=======================FINAL PROMPT======================
=========================================================
**The Architect's Deep Research Intelligence Prompt**
> Copy the entire prompt below into your preferred AI tool. Replace [bracketed text] with your custom inputs.
---
**ROLE & OBJECTIVE**
You are a senior strategic research analyst operating at the caliber of McKinsey, Gartner, or Bain. Your task is to deliver a **comprehensive, evidence-based intelligence report** on the topic defined below. Your deliverable must prioritize **verifiable data, strategic synthesis, and deep pattern recognition**. Avoid vague summaries or opinions—only include content supported by named sources or datasets.
---
**CORE RESEARCH TOPIC**
[Insert your topic here – e.g., “Go-to-market strategies for B2B SaaS,” “Customer retention in e-commerce,” etc.]
---
**INPUT CONSTRAINTS & RESEARCH BOUNDARIES**
- **Timeframe**: Limit all research to sources from the past 2–3 years only ([e.g., 2022–2025]).
- **Company Sample**: Your analysis must focus on at least [Insert number – e.g., 5] specific, named companies that are known leaders or innovators in this domain. Those companies are:
[Insert company names: Company A, Company B, Company C, etc.]
Ensure diversity in geography, company size, or growth stage where relevant.
- **Data Requirements**: For each company, synthesize **multiple types of verifiable information**, including but not limited to:
- Documented strategies, frameworks, or business model diagrams
- Executive interviews (written, audio, or video)
- Public performance metrics or internal case studies
- Third-party analysis from credible sources (e.g., TechCrunch, HBR, McKinsey, a16z, etc.)
- **Citations & Transparency**:
- Cite every major claim or data point.
- If no source can be found, state clearly: “No verifiable source found.”
- Highlight any contradictions, discrepancies, or data gaps in the source material.
---
**PRE-ANALYSIS STEP**
Before diving into data collection, generate 2–3 initial hypotheses about likely strategic themes or differentiators in this space. These will guide your research and be tested as you gather information.
---
**OUTPUT STRUCTURE**
Present your final research as a professional report with the following clearly labeled sections:
1. **Executive Summary**
- A 1-paragraph overview of your most critical findings and implications.
2. **Key Strategic Patterns (3–5)**
- What high-level strategic principles or patterns emerged across most of the companies?
- Include short illustrative examples and note any companies that diverged.
3. **Company-by-Company Deep Dive**
- For each company:
- Specific strategies/tactics
- Outcomes (metrics, market response)
- Noteworthy frameworks or quotes
- Any unique risks or tradeoffs they faced
4. **Contrarian or Non-Obvious Insights**
- Identify any findings that contradict typical “best practices” or industry assumptions.
5. **Actionable Opportunities for a New Entrant**
- What 3–5 strategic plays could a new company entering this space reasonably pursue based on your analysis?
- Justify each recommendation using prior examples or patterns.
6. **Cited Sources**
- Include a complete list of links, articles, interviews, reports, or datasets used.
---
**Formatting Tips**
- Use bullet points, bold headers, or tables where appropriate.
- Keep writing precise, structured, and insight-driven.Why Size Matters (In Research)
When you force your AI to go deep, magic happens:
Patterns emerge you’d never spot in a summary
Contradictions appear that spark original thinking
Edge cases reveal opportunities others miss
Real data replaces conventional wisdom
You’re not reading all 42 pages word-for-word. You’re mining for strategic gold. The goal isn't to read the report; it's to find the 3-5 golden nuggets that will form the core of your strategy. And trust me, there’s always gold in that depth.
The Multi-LLM Advantage
Here’s my actual workflow:
Start with Gemini for the deep dive (best at comprehensive research)
Feed key findings to Claude for critical analysis
Use ChatGPT to pressure-test assumptions
Let Perplexity verify specific claims Four different perspectives.
One strategic vision. You’re not using AI. You’re conducting an AI orchestra.
What If This Became Your Default?
Imagine walking into every meeting with strategies built on 40+ pages of verified research. Not skimmed. Not summarized. SYNTHESIZED. You become the person who brings depth in a world of surface-level thinking.
The one who says, “I analyzed 10 companies who solved this exact problem. Here’s what actually worked—with receipts.” That’s not just strategic thinking. That’s strategic authority.
Today’s Action (15 Minutes to Start, Let it Run) 🫵 💥
Time to generate your first massive research report. This is how we make the lesson real.
Pick a strategic challenge you’re currently facing.
Use the Architect's Prompt I gave you above. Fill in the brackets with your specific companies, timeframe, and data requirements.
Run the prompt in your favorite deep research tool and let it cook. Don't rush it.
Skim the final report for patterns. Your only job is to find 3-5 "golden nuggets"—recurring themes, surprising contradictions, or non-obvious insights.
That massive document you just created? Tomorrow, we’ll transform it into your first Blueprint Strategy.
We’re going to take that research and distill it into strategic dynamite.
Sam
P.S. Which LLM did you use for your deep dive? I’m genuinely curious—everyone develops their own preferences. Hit reply and let me know what you discovered in your research.
P.P.S. Save that research report. By Day 10, you’ll be shocked at how many ways you can repurpose that single document. One input, infinite outputs. That’s the engine we’re building.