Brick 1
Flow 1: SEO Content Machine
Research keywords -> scrape page-1 -> write with perspective -> publish via CMS API -> track GSC -> optimize -> batch all keywords.
Source: src/lib/bricks/sources/gtm-engineering-flows-combined.md
Brick 2
Framework-as-Diagnostic Overlay
Greg's 7-tactic framework overlaid on Yuri's current infra: | Tactic | Greg's framework says | We have | Verdict | |---|---|---|---| | 1. MCP server as sales team | Publish to Smithery/MCPT/OpenTools for $0 CAC discovery | Zero — neither BizBuilder/KPDD nor Solacian have an MCP server | **REAL GAP** — but applicability depends on whether the product answers a queryable question; Solacian (Maze-dissolving AI) plausibly does, KPDD (PMF discovery) plausibly does | | 2. Programmatic SEO at 10K-page scale | Next.js + Firecrawl + AI content for "best X for Y" patterns | Zero programmatic pages; KPDD/Solacian don't have SEO surface | **REAL GAP** for both; needs a keyword pattern decision before any code | | 3. Free tool as top of funnel | Grader/analyzer/calculator, instant value, share-driven loop | KPDD itself is arguably a discovery tool but is gated by signup, not free-tool-grade | **PARTIAL GAP** — KPDD could expose a free analyzer slice; Solacian could expose a free "Maze diagnostic" | | 4. AEO (Answer Engine Optimization) | Top-20 questions + FAQ schema + monitoring via Otterly/Profound | Zero structured FAQ surface on either product site | **REAL GAP** for bo…
Source: src/lib/bricks/sources/greg-isenberg-bootstrap-distribution.md
Brick 3
Strategy 2: Programmatic SEO (10,000 Pages)
The thesis: create 10,000 SEO pages in 48 hours via keyword patterns + structured data + AI-generated unique content. Math model: - 10,000 pages × 30 visits/month each = 300,000 monthly visitors - 2% conversion = 6,000 conversions/month - $10 each = **$60,000/month from pages built once** - Caveat: 30 visits/month doesn't happen overnight; SEO compounds over time Critical: content must not feel like AI. Lots of optimization needed. Start with a few pages, scale once quality is right. "Press one button" myth is rejected explicitly. Playbook (start this week): 1. Pick a keyword pattern: `[product type] for [niche]` or `[service] in [city]`. Examples: "CRM for dentists", "roofing in Miami" 2. Build dataset via Firecrawl (scrapes + clean structured data) or existing databases 3. Create page template in your framework: Next.js, Webflow, WordPress — pick what you use 4. Use AI to generate unique paragraphs per page (not variable swaps; high-quality content) 5. Human-in-the-loop editing on a sample 6. Publish 100 pages as MVP 7. Monitor indexation 8. Scale once indexed Applies to: services, SaaS, apps, boring businesses, agencies. Programmatic SEO is still under-tapped.
Source: src/lib/bricks/sources/greg-isenberg-bootstrap-distribution.md
Brick 4
Flow 6: Content Improvement Loop
Graph MCP -> GSC -> find underperforming pages -> identify striking distance keywords (position 5-20) -> Claude rewrites sections -> republish -> track.
Source: src/lib/bricks/sources/gtm-engineering-flows-combined.md
Brick 5
1.4 Autonomous Ad Optimization Loop (Cody method)
Source: [[../knowledge-env/synthesized/gtm-engineering-flows-combined.md]] **When to use**: running 50+ ad variations, need daily kill/promote cycle without manual monitoring **Process**: 1. Test campaign runs new creative constantly (bulk-generated) 2. Daily cron job: pull performance data from data warehouse 3. Agent identifies high-CPM/low-performing ads 4. Auto-pause losers via Facebook Ads API 5. High performers get promoted to dedicated ad set with own CPA budget 6. Dashboard tracks the whole loop (clicks over time, CPC, demographics) 7. Morning brief via MCP: "How many new users went to homepage yesterday?" **Key insight**: Direct Facebook Ads API MCP has a pagination problem (you only see ~5% of your data). Need a data pipeline + data warehouse (like Graph) to get the full picture at scale. **Deployment progression**: - Start: manual in Claude Code terminal - Scale: deploy to Railway as always-on agent with cron schedule - Team: deploy UI to Vercel for team access - On-the-fly: spin up ephemeral Postgres on Railway for ad-hoc data analysis, spin down when done **Tools**: Claude Code, Facebook Ads API, Railway (deployment + ephemeral DB), Graph MCP (…
Source: src/lib/bricks/sources/perf-marketing-playbook.md
Brick 6
1.5 Competitor Creative Mining + AI Tagging (Apify pipeline)
Source: [[competitor-creative-mining-apify-pipeline.md]] **When to use**: entering a niche, no creative ideas, need to know what's working in competitor ads RIGHT NOW. Pre-step before §1.3 (code generation) — gives you the parameter library to remix from. **Process**: 1. Apify actor `curious_coder/facebook-ads-library-scraper` ($0.20-0.75 / 1K ads) pulls competitor ads from Meta Ads Library by search URL or competitor page list 2. Edit node computes `days_running = (end_date - start_date) / 86400` — proxy for performance (winners run 30+ days, losers killed in <7) 3. Download creatives to Google Drive or Supabase 4. GPT-4o vision analyzes each image, returns structured JSON: `ai_hook`, `ai_offer`, `ai_cta`, `ai_psychology`, `ai_utp`, `ai_jtbd`, `main_object`, `main_color`, `ai_description` 5. Code node validates JSON schema (one column = one property) 6. Merge node joins AI tags with metadata 7. Google Sheets append: one row per ad with image, params, days running, reach, est. spend 8. **Spend hack**: `est_spend = reach × avg_CPM_in_geo`; rank by `est_spend × days_running` to find competitor hero creatives 9. Output sheet feeds downstream creative-generation a…
Source: src/lib/bricks/sources/perf-marketing-playbook.md
Brick 7
B041 — Interactive tool (SEO audit, ROI calculator, readiness quiz) vs PDF
Interactive tool (SEO audit, ROI calculator, readiness quiz) vs PDF. Higher perceived value, email capture to view results. **Tool candidates:** T001 Claude Code + T116 Typeform or custom build.
Source: _reference/bricks/README.md