Example: Content Creation Workflow

A complete walkthrough of using Digital Brain for content creation.

Scenario

You want to write a Twitter thread about "lessons from building in public."

Step 1: Agent Activates Skill

User Input: "Help me write a thread about lessons from building in public"

Agent Recognition:

• Trigger: "write a thread" → content creation
• Topic: "building in public" → likely matches brand pillars
• Action: Load Digital Brain skill

Step 2: Load Identity (Required)

Agent reads identity/voice.md:

# Extracted patterns:
formal_casual: 7/10 (fairly casual)
signature_phrases:
  - "Here's the thing"
  - "Let me break it down"
power_words:
  - "ship", "iterate", "leverage"
avoid:
  - "synergy", "circle back"
  - excessive emojis

Agent reads identity/brand.md:

# Confirms topic alignment:
content_pillars:
  - building_in_public  ✓ MATCH
  - ai_agents
  - founder_lessons
target_audience: "Early-stage technical founders"

Step 3: Reference Past Success

Agent scans content/posts.jsonl:

// Found similar high-performing thread:
{
  "type": "thread",
  "pillar": "building_in_public",
  "metrics": {"likes": 234, "reposts": 45},
  "notes": "Story format worked well"
}

Insight: Story-driven threads perform best for this pillar.

Step 4: Check Ideas Bank

Agent scans content/ideas.jsonl:

// Found relevant undeveloped idea:
{
  "idea": "Thread on the emotional rollercoaster of public building",
  "status": "raw",
  "priority": "high"
}

Decision: Can incorporate this angle into the thread.

Step 5: Generate Draft

Agent uses content/templates/thread.md as scaffold:

## 1/ Hook
Unpopular opinion: Building in public isn't about the wins.

Here's the thing - it's about the messy middle.

Let me break it down 🧵

## 2/ Context
I've been building in public for 18 months.

Here's what I wish someone told me on day 1:

## 3-7/ Main Points
[5 specific lessons with stories]

## 8/ Takeaway
The compound effect of transparency beats any marketing strategy.

## 9/ CTA
Follow for more founder lessons as I ship @handle

Voice Alignment Check:

• ✓ Uses "Here's the thing" (signature phrase)
• ✓ Uses "Let me break it down" (signature phrase)
• ✓ Casual but substantive tone (7/10)
• ✓ No corporate jargon
• ⚠️ Limited emoji (one 🧵 - acceptable)

Step 6: User Feedback & Iteration

User: "Can you make point 3 about the fear of judgment?"

Agent:

1. Keeps same voice/structure
2. Rewrites point 3 with fear angle
3. Maintains thread flow

Step 7: Log the Idea (If Not Publishing)

If user saves for later, agent appends to content/ideas.jsonl:

{
  "id": "idea_20241229_160000",
  "created": "2024-12-29T16:00:00Z",
  "idea": "Thread: 5 lessons from building in public (fear, wins, community...)",
  "source": "developed_draft",
  "pillar": "building_in_public",
  "status": "ready",
  "priority": "high",
  "notes": "Draft complete, reviewed voice alignment",
  "tags": ["thread", "building_in_public", "founder_lessons"]
}

Step 8: Post-Publish Logging

After user publishes, agent appends to content/posts.jsonl:

{
  "id": "post_20241229_180000",
  "published": "2024-12-29T18:00:00Z",
  "platform": "twitter",
  "type": "thread",
  "content": "Unpopular opinion: Building in public isn't about the wins...",
  "url": "https://twitter.com/user/status/123456789",
  "pillar": "building_in_public",
  "metrics": {
    "impressions": 0,
    "likes": 0,
    "comments": 0,
    "reposts": 0
  },
  "metrics_updated": "2024-12-29T18:00:00Z",
  "notes": "Story-driven format, fear angle resonated in drafting",
  "tags": ["thread", "building_in_public"]
}

Files Accessed

Total: ~650 tokens vs loading entire brain (~5000 tokens)

Key Takeaways

1. Voice First: Always loaded before drafting
2. Progressive Loading: Only relevant modules accessed
3. Pattern Matching: Past success informs new content
4. Full Pipeline: Idea → Draft → Publish → Log
5. Append-Only: Ideas and posts logged, never deleted