💎 Why distill knowledge
The rule that guides everything: generating text is cheap; space in the context window is expensive. Distilling means trading volume for density. Every token that goes in has to pay a relevance tax—or it’s taking up space that could be used for something better.
💡 The trade-off that changes everything
It’s not about having less knowledge—it’s about having the same knowledge take up less space. A 1-page cheat sheet can carry the essentials of a 46-page PDF. The difference directly affects the quality of Claude Code’s response.
is cheap
is expensive
> volume
pays for the spot
🔎 Curate Sources
Garbage in, garbage out. The quality of your foundation depends on the quality of the source — so aim for authority. McKinsey, BCG, KPMG, OpenAI, and IBM publish free guides to showcase their expertise. You get elite knowledge for free.
🏛️ BCG & McKinsey
Maturity curve, "where's the value in AI," use cases by sector. Frameworks that become deliverables.
📋 KPMG
"Agentic AI advantage" — the shift toward agents, governance, and change management.
🤖 OpenAI
CEO's guide to generative AI: identifying and scaling use cases. A practical implementation overview.
📊 Google Cloud & IBM
AI ROI with real numbers, state of gen AI. Statistical ammunition for ROI.
💡 Where to find it for free
Search for “[consulting] AI report PDF” or go straight to the section on insights / research from their website. These are reports designed to generate leads—that is, intentionally published. Download, distill (module 2.2), and archive them in your knowledge base.
elite source
public guides
you choose
inherited rigor
🧬 Anatomy of a reusable cheat sheet
Consistent structure is what makes the foundation searchable and injectable. When every sheet follows the same template, each one becomes a plug-in piece—and you (or Claude Code) can find what you need at a glance.
Concepts & Frameworks
The core models, each in 1–2 sentences. The outline of what the source teaches.
Critical statistics
Numbers, percentages, and findings—with the source. That’s the ammunition that gives the deliverable weight.
Insights + Citations + Terms
Actionable recommendations, 2-3 impact citations, and a mini-glossary of acronyms. Always scannable, with the term in bold.
same template
bullets
in the term
by source
🗂️ Version and organize the knowledge base
Curated knowledge is an asset that grows in value over time— if you don't lose control of it. Organize the folder by topic, date each sheet, record the source, and version it with git. The knowledge base evolves without becoming a mess.
// a structure that scales
conhecimento/ ├── maturidade/ │ └── bcg-maturity-curve_TLDR.md ├── roi/ │ ├── google-cloud-roi-2025_TLDR.md │ └── ibm-state-of-genai_TLDR.md ├── governanca/ │ └── kpmg-agentic-advantage_TLDR.md └── README.md # índice: o que é cada folha + fonte + data # versionar: git add . && git commit -m "add ROI cheat sheets"
✓ Foundation under control
- ✓Folders by topic
- ✓Each page has a source and date
- ✓README as an index
- ✓Version-controlled with Git
✗ Knowledge base turning into a mess
- ✗50 loose files in a folder
- ✗Without knowing where each one came from
- ✗Duplicate and outdated versions
- ✗Names like "final doc v3 FINAL"
clear folders
traceable
index
versions
💉 Inject context without exceeding the token limit
It’s lazy RAG applied to everyday work. Instead of dumping the entire knowledge base, you pull only the relevant cheat sheet for the task at hand. The human does the retrieval — on demand, from memory.
✗ Dump everything
- ✗"Read the entire conhecimento/ folder"
- ✗Context window fills up, response slows down
- ✗What matters gets lost in the middle
✓ Pull out what matters
- ✓"Use the Google Cloud ROI cheat sheet"
- ✓Lean context window, sharp response
- ✓The right context > a large context
// what you type to Claude Code
Generate the 30/60/90 roadmap for “Clínica Vida.” Use only these two cheat sheets as references: conhecimento/maturidade/bcg-maturity-curve_TLDR.md conhecimento/governanca/kpmg-agentic-advantage_TLDR.md
applied
manual
only what's relevant
lightweight window
📖 Your Personal Consultant Library
Together, they form the final asset: the folder conhecimento/ consolidated. Several curated cheat sheets that power the entire Factory and go with you from project to project. Built once, used for every client.
🧠 This is what makes the Factory sound like an elite firm
Remember module 1.2? The output looks like BCG/KPMG consulting because the prompts were fed this distilled knowledge. Your knowledge base é the machine's brain.
- •Feeds the 15 deliverable prompts (Track 4)
- •Becomes a reference attached to your skills (Track 3)
- •Grows with every project — compounding advantage
💡 Start small, always compose
You don’t need 50 cheat sheets to get started—3 well-curated ones can already change the game. With each new client, distill one more relevant source. In a year, you’ll have a library no competitor can improvise.
every project
the Factory
stays with you
always grows
✅ Module summary
🎯 Mission 2.3 — Your conhecimento/ folder
Build the seed of your consultant library:
- Create the folder
conhecimento/with subfolders by topic. - Set up 3 curated cheat sheets (use the ones already distilled or generate new ones with module 2.2).
- Write a
README.mdindex: name, source, and date of each sheet. Version it with git.
Success: folder with 3 cheat sheets + versioned README. What you gained: the consultant's brain that will power your entire Factory—and grows with every project.
Next track:
Track 3 — Skills & Agents: turn this knowledge into reusable resources that Claude Code launches on demand.