Built with AI

POST /classify-reply
// Response:
{
  "sentiment": "positive_interest",
  "confidence": 0.95,
  "draft_reply": "Hey John..."
}

# Multi-layer filtering
if is_own_domain(email):
    return skip
if is_ooo(body):
    return "automated_ooo"
if is_duplicate(lead, 24h):
    return skip

POST /api
{"action": "trigger"}

// Pipeline:
Cal.com → Research → Word Doc → Slack
// Cron: hourly on weekdays

// 7 deliverables in one call:
→ Press release (400-700 words)
→ FAQ + FAQPage schema
→ JSON-LD (Org, Product, HowTo)
→ robots.txt for AI crawlers
→ llms.txt specification
→ Directory copy
→ Monitoring prompts

POST /api
{"action": "run_pipeline"}

// Autonomous flow:
Keywords → Plan → Generate → Publish
// Direct to WordPress

# 4 apps × 7 channels
newsletter-source   → Aggregate
newsletter-curate   → Filter
newsletter-generate → Write + Images
newsletter-publish  → 7 platforms

POST /generate-content
{"idea": "AI is transforming..."}

// Returns:
→ LinkedIn post (150-300 words)
→ Twitter thread (5-7 tweets)
→ Beehiiv HTML newsletter

# Smart column mapping
"email_address" → email
"phone_number" → phone
"company"     → company_name
# 30+ aliases auto-mapped

POST /generate-proposal
// Pipeline:
Fathom Call → Transcript
  → Claude Analysis
  → Google Docs
  → Auto-share link

POST /webhook  # LOGIN|CHAT|LOGOUT
POST /transcript-sync
POST /analyze   # Claude pipeline
GET  /results   # Fetch analysis

# WAT Framework
workflows/
  → persona-research.md
  → content-calendar.md
  → campaign-launch.md
tools/
  → competitor_analysis.py

# B2C Marketing
Target: Parents
Channels:
  → Social (Instagram, FB)
  → Content Marketing
  → Community Building

# Infrastructure stack
Domains → Warm-up → Sequences
  → Send → Track → Classify
  → Auto-reply / Escalate

# Lead scoring
for post in subreddit.search():
    score = intent_score(post)
    if score > threshold:
        leads.append(post.author)

# Signal triggers
→ New job posting (hiring)
→ Funding round announced
→ Tech stack change detected
→ Auto-trigger outreach

# Pipeline
YouTube video
  → Extract audio
  → Generate artwork (Modal)
  → RSS feed
  → Podcast directories

# Recording pipeline
Riverside.fm export
  → Process video
  → Generate thumbnail
  → Format metadata
  → Upload to YouTube

# Content transformation
Videos → Transcribe
  → Structure chapters
  → Edit & format
  → Generate manuscript

# One idea → many platforms
idea = "AI automation tip"
→ LinkedIn (long-form)
→ Twitter (thread)
→ Instagram (carousel)
→ Newsletter (deep-dive)

# Outreach flow
Research podcasts
  → Score relevance
  → Craft pitch (Claude)
  → Send & follow up
  → Track bookings

# 4 Modal endpoints
marygrove-script-gen
marygrove-script-revise
marygrove-validate-video
marygrove-credit-tracker

# LinkedIn automation
Content calendar
  → Generate posts (Claude)
  → Schedule publishing
  → Track engagement

# Agency automation
Campaign setup
  → Creative generation
  → A/B test variants
  → Performance reports

# MCP Server
@mcp.tool()
def read_notes():
    # Bridge Apple Notes → Claude
    return apple_notes.fetch()

# Knowledge bridge
Obsidian vault
  → Claude Code context
  → Research & synthesise
  → Write back to vault

# Telegram bot
@bot.command("trigger")
async def handle(msg):
    result = run_workflow()
    await bot.send(result)

# Track AI ecosystem
Sources:
  → Product Hunt launches
  → GitHub trending
  → Funding announcements
  → Research papers

# LinkedIn automation
Connect → Engage → Convert
  → Profile visits
  → Content likes
  → DM sequences

# Course modules
Module 1: Foundations
Module 2: Agents & Tools
// Case study:
Basecamp Coffee loyalty program

# Slash commands
/prime    → Set up workspace
/create-plan → Strategy doc
/implement  → Execute tasks
# Learn by doing

# Template structure
CLAUDE.md        # Instructions
workflows/       # SOPs
tools/           # Python scripts
.claude/commands # Slash cmds

# Key patterns taught
1. Ask User Question → Align
2. Slop in / Slop out
3. Fathom → Proposal pipeline
# Live demo workflow

# 3 autonomous endpoints
reply-classifier  → Triage
email-copywriter  → Personalise
niche-scorer      → Prioritise
# = Autonomous sales

# Scoring criteria
market_size      × 0.25
competition      × 0.20
ai_readiness     × 0.30
monetisation     × 0.25
# → Ranked opportunities

<!-- Bertie brand site -->
<header>
  <nav>...</nav>
  <hero>...</hero>
</header>

# Video pipeline
Script (Claude)
  → Storyboard
  → Generate scenes
  → Assemble promo

# 90-day roadmap
Days 1-30:  Foundation
Days 31-60: Growth
Days 61-90: Scale
# KPIs tracked weekly