Jonathon's AI Wiki

AI Marketing

9 min read

AI-powered marketing workflows — campaigns, automation, content strategy, ad creative, SEO, analytics, and personalization. Focus is on practical application of AI tools (especially Claude) to marketing problems.

Articles

  • Claude Cowork for Marketing — Using Claude’s desktop agentic mode for marketing tasks: strategy decks, ad creatives, campaign analysis, and scheduled automation.
  • Cowork + Apify Scraping Recipe — No-code 5-minute Cowork recipe (Notion AI Recipe, March 2026): connect Apify connector, then chain Cowork prompts to scrape local-business directories → “vibe prospecting” on individuals at those companies → LinkedIn job-description scrape → PR-ready research report on AI-skill demand. Five concrete deliverables in one session. Pairs with Clawdbot (structured-pipeline competitor scraper) as the chat-first ad-hoc alternative. Source has step contents inside Notion toggles that didn’t render in the static-HTML extract — outline + tips + forward applications captured; in-toggle screen-by-screen walkthroughs left to the source.
  • The LinkedIn Engagement Machine — Four-prompt-chain LinkedIn content workflow (Notion AI Recipe, January 2026): raw video transcript or voice notes → 3-5 strategic post concepts → fully-drafted post → ruthlessly-edited final post → optional lead-magnet outline. Same-thread context-retention is the architectural choice — all prompts run in one chat so the model retains transcript context end-to-end. Anchored on the Justin Welsh persona (parenthetical proof, pattern interrupt, confession, timestamp credibility). Source claims 3-5x more comments than generic AI posts and 20-50+ lead-magnet requests per post. Includes batch-the-ideation, repurpose-everything, and save-as-Project pro tips.
  • AI Marketing Automation Use Cases — Concrete use cases for Claude in email marketing, competitive intelligence, personalized outreach, ecommerce, and lead generation.
  • Outcome Kit — The AI Agent That Knows Which Ads Actually Print Money — Matthew Berman’s open-source 3-agent pipeline mapping Meta Ads + GA4 + Calendly/HubSpot to the Four Outcome Truths (best/fake/lurking/ungraded winners). MIT, runs on Claude Code / OpenClaw / Hermes.
  • Stanford HAI AI Index 2026 — Marketing Cuts — The marketer’s extract from Stanford HAI’s 423-page flagship report. Adoption stats (53% pop / 88% org / 80% student), $172B/yr consumer value, 14-26% productivity gains, the 73/23 trust gap, jagged-frontier framing, and 5 ready-made slide templates citing CC BY-ND 4.0 data. Companion to the full summary.
  • 2026 Business Demand for AI Workflows — Field data from 500+ deployed AI workflows ranking the categories that reliably print revenue (speed-to-lead, document processing, follow-up sequences, database reactivation, internal reporting), pricing signals from the dental / accounting / webinar / gym / construction-crew worked examples, and the recurring failure modes (“clients want AI agents but their actual clog is upstream”). Plus the diagnostic question: if 500 new clients showed up tomorrow, what would break first?
  • The 5 AI Automations Businesses Actually Pay For — Field-tested catalog of the five highest-paying AI automation patterns from the AIS+ resource library. (1) Speed-to-Lead (auto-respond to inbound leads inside ~5 minutes — every minute past 5 cuts conversion 80%), (2) Document Processing (intake forms, invoices, contracts → structured data → CRM), (3) Follow-Up Sequences (multi-touch nurture pipeline that the agent owns end-to-end), (4) Database Reactivation (mining the “dead leads” CRM segment for cold-but-real intent), (5) Internal Reporting (collapsing weekly Monday-morning report-building into pre-generated dashboards). Each pattern includes ROI math worked through with concrete numbers, the Pipe Analogy for diagnosing whether the bottleneck is in the right place, and the diagnostic question “If 500 new clients showed up tomorrow, what would break first?” as the universal lead-in prompt. Companion to 2026 Business Demand for AI Workflows (covers the same five categories with field data) — this article is the implementation-pattern view, that one is the demand-side view. Sister course to Codex 1-Hour / Paperclip / Hermes 1-Hour in the AIS+ resource bundle.
  • How Alex Hormozi Uses AI — The 4-Pillar System — Nate’s walkthrough (45JqVBihguo) of Alex Hormozi’s AI operating system as reverse-engineered from interviews, job postings, social media, and Leila Hormozi’s internal anti-AI-slop memo at acquisition.com. Four pillars: (1) Your data is your moat — Hormozi built “Ac AI” trained on 31M-dataset users is export tweets/emails/sales calls + add Whisper-Flow/Hex voice-capture + meeting transcribers, then prompt “Based on everything I’ve uploaded… reference what’s worked before.” (2) Voice-first, AI second — Leila’s memo: “I am so sick of reading AI slop, especially in memos” (tells: “delve into,” “this signals that,” “synergies”); use voice dictation for first draft, AI to make it better not to replace your thinking. Prompt pattern: “Interview me to help me identify any gaps in my thinking.” (3) Break up your workflows — role-based thinking → workflow-based thinking; a prompt IS a Standard Operating Procedure. (4) The compound effect — amplification, not automation — realistic numbers (Hormozi published): emails 4x throughput, tweets 7x, YouTube 8x. Pillar 4 is what Pillar 1 turns into over time. Hormozi’s “18 months where there’s just a huge amount of wealth that can be created by people who have nothing” framing.
  • illo — Editorial-Illustration Agent Skill with a Recurring Mascot — tmchow/illo-skill (Trevin Chow, MIT, 90★; the GitHub-Trending auto-caption heard “illo” as “ELO”). Turns an idea or a whole article into an original print-style editorial illustration starring a recurring mascot — a consistent brand house style instead of stock photos or generic AI slop. Ten bundled looks (riso default + blueprint/woodcut/pixel/clay/manila/chalk/phosphor/enamel/gouache) + a character builder; two registers (editorial scene + hand-built explainer diagram) + mini-comics. Dual rendering backend = a real cost lever: your Codex CLI (gpt-image-2 on a Codex subscription — $0/image, no API key) or OpenRouter (Grok Imagine / Nano Banana / GPT-5.4 Image 2) as the universal fallback. Disciplined trigger (fires only on explicit “illo”), mode-600 key hygiene, and installs across ~70 agent runtimes via skills.sh. The anti-AI-slop thesis of Refero / Hormozi’s “no AI slop” applied to article art; same gpt-image-2-via-Codex pipeline as the team’s codex-imagegen. Competitor: orange2ai/orange-line-illustration (135★, commercial license for closed-source). Flagged in the GitHub-Trending Weekly 36 roundup.
  • Lead Magnet Creation with Claude Code (Brandon Storey course) — Brandon Storey’s nearly-two-hour Copywriting Coach walkthrough of an end-to-end lead-magnet pipeline. Six sequential phases — ideation (Claude web + community-question dumps + Fathom transcripts) → production (Claude Code with a downloaded design skill + brand-guidelines-found-on-disk) → triple lead magnet (Doc + PDF + custom GPT) → email automation (Beehiiv) → landing/thank-you pages (Kajabi) → social distribution (Vizard captions + Twitter post + ManyChat Instagram comment-to-DM). Glued with Zapier. CTA-at-every-stage discipline pulls people from free download into the paid Six Figure Copy Academy from the first touch. The “AI copywriter as data synthesizer” framing (extract → push → refine) is the underlying mental model. Editorial caveat: video itself is a lead magnet for the speaker’s paid academy.

Sales & Client Acquisition

  • AI Automation Client Acquisition Playbook — How AI service agencies / freelancers land their first clients. The cold-email framework (subject lines, body templates, send volume, day-of-week timing), lead-identification signals, the proof-driven close pattern (showing real working automations vs pitching hypotheticals), the 7-day timeline, and the warm-then-cold-then-Trojan-horse channel sequence for founders with zero followers. WEO Marketly translation: same playbook adapts to landing dental / healthcare practice clients.
  • Karpathy-Style AutoResearch for Cold Outbound — A B2B lead-gen agency applies Karpathy’s AutoResearch experiment-loop idea to cold email, optimizing positive reply rate. Deep context engineering (ICP / case-study / value-prop / problem-statement MD files from the site + a voice-memo transcript) + a pre-built full-TAM enrichment (Clay / Apify / Prospeo / Rapid API) so the agent makes no “game-time decisions”; “the list is the message” generates campaign experiments with 3–5 verified contacts each; locked-CTA + Million Verifier + hard-coded human-approval guardrails; loads into SmartLead / Instantly with a weekly Claude Code routine / Codex automation proposing the next round. Reports an enterprise client’s reply rate doubling (20.71 vs 10.71 replies/1k, sustained). Applied generalization of Karpathy’s AutoResearch.
  • Meta Ads CLI — Command-Line Interface for Meta Ads and Commerce — Meta launched the official Ads CLI on April 29, 2026. Python 3.12+ via pip/uv. Packages the Marketing API into one tool with predictable commands for both developers and AI agents — covers campaigns, adsets, ads, creatives, insights, catalogs, products, conversion-pixel datasets. Three output formats (table / json / plain), standard exit codes, env-var token support, resources created in PAUSED status by default (key safety affordance for agent-driven campaign staging). Vendor-shipped agent tooling, mirroring the Railway Remote MCP pattern.
  • OpenAI Ads — Advertising Inside ChatGPT (Ads Manager Beta) — OpenAI launched a self-serve advertising platform that places paid ads below relevant ChatGPT conversations. Free and Go tiers only (Plus/Pro/Business are ad-free); US/CA/AU/NZ only; CPM and CPC pricing on a relevance-weighted second-price auction. The Ads Manager schema mirrors Google Ads structure (Campaign → Ad Group → Ad) for near-zero learning curve, with a Marketing API at developers.openai.com/ads.
  • Similarweb — Ads in AI: Insights from Real User Behaviour — First cross-platform performance study of paid AI advertising, covering ChatGPT, Google AI Mode, and AI Overviews side by side (Similarweb panel data, Mar 30–Apr 13 2026). Conversational ads behave fundamentally differently from search ads: ChatGPT reads the full conversation rather than matching keywords, ads fire later in the session, and 46% of users who opened with zero commercial intent developed buying signals by the time an ad appeared. The empirical backbone for the OpenAI Ads launch.

Social Media

  • Postiz — Open-Source AI Social Media Scheduling Platform — 29.2k-star open-source social-media scheduler (gitroomhq/postiz-app). 14 networks (Instagram / YouTube / LinkedIn / X / Threads / TikTok / Bluesky / Mastodon / Slack / Discord / Reddit / Pinterest / Facebook / Dribbble), AGPL-3.0, fully self-hostable. NextJS + NestJS + Prisma + Temporal monorepo. Native AI content generation, AI scheduling, Postiz Agent CLI for autonomous-agent integration, plus N8N node / Make / Zapier / NodeJS SDK. Open-source answer to Buffer / Hootsuite / Later.

17 items under this folder.