Skip to main content
This guide assumes you’ve already completed the installation and setup wizard. If you haven’t, run npx mega-brain-ai setup first.

Overview

Mega Brain transforms expert materials into structured knowledge through a 5-phase processing pipeline. In this guide, you’ll:
  1. Ingest your first material (video, PDF, or transcript)
  2. Process it through the JARVIS pipeline
  3. View the generated knowledge artifacts
  4. Check your system status
Make sure your OPENAI_API_KEY is configured in .env before proceeding. Video and audio ingestion requires OpenAI’s Whisper API for transcription.

Step 1: Ingest Your First Material

Mega Brain accepts any expert material - videos, PDFs, transcripts, podcasts, or training courses. Let’s start with a YouTube video.

Using Claude Code

Open your Mega Brain project in Claude Code and use the /ingest slash command:
1

Open Claude Code

Launch Claude Code and open your Mega Brain project directory.
2

Run the ingest command

In the Claude Code chat, type:
/ingest https://www.youtube.com/watch?v=EXAMPLE_ID
Replace EXAMPLE_ID with your actual YouTube video ID.
3

Review the output

JARVIS will download and process the transcription:
JARVIS: Material received.

  Source:    YouTube
  Title:     "How to Create Irresistible Offers"
  Author:    Alex Hormozi
  Duration:  42:15
  Words:     6,230

  Saved in: inbox/alex-hormozi/how-to-create-irresistible-offers.md

  Next step: execute /process-jarvis to process.

Alternative: Ingest Local Files

You can also ingest PDFs, Word documents, or text files:
/ingest https://www.youtube.com/watch?v=dQw4w9WgXcQ
Materials are automatically organized by author in the inbox/ directory. JARVIS extracts metadata like title, author, and date from the source.

Step 2: Process with JARVIS

Now that your material is ingested, run it through the 5-phase processing pipeline:
1

Launch the pipeline

In Claude Code, execute:
/process-jarvis
2

Monitor the phases

JARVIS processes your material through 5 phases:
JARVIS: Pipeline initiated.

  Phase 1/5 - Chunking .............. OK (23 chunks)
  Phase 2/5 - Resolution ............ OK (8 entities)
  Phase 3/5 - Extraction ............ OK (12 insights)
  Phase 4/5 - Synthesis ............. OK (3 narratives)
  Phase 5/5 - Compilation ........... OK

  Result:
    Dossier:   knowledge/dossiers/alex-hormozi.md
    Playbook:  knowledge/playbooks/irresistible-offers.md
    Insights:  12 new insights indexed

  Time total: 2m 34s
3

Understand the pipeline

Each phase transforms your content:
PhasePurposeOutput
ChunkingBreak content into semantic segmentsartifacts/chunks/
ResolutionIdentify people, concepts, entitiesartifacts/insights/
ExtractionExtract frameworks and heuristicsartifacts/insights/
SynthesisConsolidate narratives by themeartifacts/narratives/
CompilationGenerate dossiers and playbooksknowledge/

Step 3: Explore Your Knowledge Base

After processing, your knowledge base contains structured artifacts:

View System Status

Check the health of your Mega Brain system:
/jarvis-briefing
Expected output:
JARVIS: Briefing Operacional

  Health Score: 92/100

  Knowledge Base:
    Experts:      1
    Dossiers:     1
    Playbooks:    1
    Insights:     12

  Last Processing:
    "How to Create Irresistible Offers" (Alex Hormozi)
    Processed on: 2026-03-06 14:30

  Active Agents:
    CRO .... Ready (12 insights loaded)
    CFO .... Ready (4 insights loaded)
    CMO .... Awaiting material
    COO .... Awaiting material

  Inbox pending: 0 files

Dossiers

Expert-focused compilations containing all knowledge from a specific person.Location: knowledge/dossiers/persons/

Playbooks

Theme-based compilations organized by topic (sales, marketing, operations).Location: knowledge/playbooks/

DNA Schemas

5-layer cognitive models extracted from expert materials.Location: knowledge/dna/

Insights

Individual extracted learnings with source traceability.Location: artifacts/insights/

Step 4: Extract DNA (Optional)

For a deeper understanding of an expert’s thinking, extract their cognitive DNA:
DNA extraction works best with 3-5+ materials from the same expert. With only one material, the DNA will be limited but still useful.
/extract-dna alex-hormozi
Expected output:
JARVIS: DNA extraction initiated.

  Expert: Alex Hormozi
  Materials analyzed: 1

  Layer L1 - Philosophies .......... OK (5 beliefs extracted)
  Layer L2 - Mental Models ......... OK (3 frameworks)
  Layer L3 - Heuristics ............. OK (8 practical rules)
  Layer L4 - Frameworks ............. OK (2 methodologies)
  Layer L5 - Methodologies .......... OK (1 complete process)

  DNA saved in: knowledge/dna/alex-hormozi-dna.md
  Agent created: agents/persons/alex-hormozi.md

  Now you can consult:
    /ask alex-hormozi "How should I price my high-ticket offer?"

The 5 DNA Layers

1

L1 - Philosophies

Core beliefs and worldviewExample: “Value is created in the gap between cost and perceived benefit”
2

L2 - Mental Models

Thinking and decision frameworksExample: “The Value Equation (Dream Outcome + Perceived Likelihood - Time Delay - Effort & Sacrifice)”
3

L3 - Heuristics

Practical rules and decision shortcutsExample: “Always sell the result, not the mechanism”
4

L4 - Frameworks

Structured methodologiesExample: “The Grand Slam Offer Framework”
5

L5 - Methodologies

Step-by-step implementationsExample: “7-Step Offer Creation Process”

Step 5: Query Your Knowledge

Once you have processed materials, you can query your knowledge base:

Ask an Expert Agent

Query a mind-clone agent that thinks like the expert:
/ask alex-hormozi "How should I structure my offer to maximize perceived value?"

Run a Conclave Session

Get multi-perspective analysis from your council of advisors:
/conclave "Should I launch my product at $2,997 or $4,997?"
Expected output:
CONCLAVE: Session initiated.

  Question: "Should I launch my product at $2,997 or $4,997?"

  ── METHODICAL CRITIC ──
  "Before deciding on price, we need to validate the Value Equation
  (Hormozi). What's the dream outcome, time delay, effort, and sacrifice
  of your product? Price should be a consequence of perceived value,
  not an arbitrary decision. If perceived value is 10x the price,
  both options work. If not, neither works."

  Sources: knowledge/playbooks/irresistible-offers.md

  ── DEVIL'S ADVOCATE ──
  "$4,997 seems attractive for margins, but consider: what's your
  current volume of qualified leads? If conversion drops from 8% to 4%
  at the higher price, you need double the leads. Can your funnel
  handle that? What's your current CAC? Risk: high ticket + weak
  funnel = zero revenue."

  Sources: CFO unit economics analysis

  ── SYNTHESIZER ──
  "Recommendation: Launch at $2,997 with an upgrade path to $4,997.
  This allows:
  1. Validate conversion at the lower price
  2. Create case studies with first customers
  3. Raise price with social proof
  Suggested timeline: 90 days at lower price, evaluate metrics, decide."

  Consensus: 2/3 recommend starting at $2,997
  Confidence: 78% (limited by 1-expert knowledge base)

Common Workflows

Batch Ingest Multiple Materials

Process multiple materials from the same expert:
/ingest https://www.youtube.com/watch?v=VIDEO_1
/ingest https://www.youtube.com/watch?v=VIDEO_2
/ingest /path/to/expert-book.pdf
/process-jarvis

Build a Multi-Expert Knowledge Base

Diversify your knowledge with multiple experts:
/ingest https://www.youtube.com/watch?v=EXPERT_A_VIDEO
/ingest https://www.youtube.com/watch?v=EXPERT_B_VIDEO
/ingest https://www.youtube.com/watch?v=EXPERT_C_VIDEO
/process-jarvis

Save and Resume Sessions

/save

Troubleshooting

Ingestion Failed

If /ingest fails:
  1. Check your OpenAI API key in .env
  2. Verify the URL is accessible
  3. Check the logs in logs/sessions/

Processing Stuck

If /process-jarvis hangs:
  1. Check Claude Code is still responding
  2. Review logs/sessions/ for error messages
  3. Try processing again with /process-jarvis

No Insights Extracted

If the pipeline completes but extracts no insights:
  1. Check material quality - ensure the content is substantive
  2. Review the source in inbox/ - verify it was transcribed correctly
  3. Try different material - some sources work better than others

Next Steps

Configuration Guide

Configure additional API keys for enhanced features like semantic search and voice interaction

Available Commands

Explore all available slash commands and CLI tools

DNA Schema Guide

Deep dive into the 5-layer DNA extraction system

Agent System

Learn about Cargo Agents (CRO, CFO, CMO, COO) and Mind Clones
Pro Tip: The more materials you ingest from the same expert, the richer your DNA extraction becomes. Aim for 3-5 materials per expert for optimal results.

Build docs developers (and LLMs) love