AI Marketing OS
varunk130/ai-marketing-os
Time de IA para cada função de marketing — skills + agents + orquestração
Ver no GitHubSistema operacional para marketing baseado em 9 agentes IA especializados e 12 skills de metodologia, que orquestra workflows end-to-end para gerar artefatos reais (listas de leads, planos A/B, relatórios de ROI). Substitui stacks de ferramentas caras por um time IA coordenado, eliminando silos de expertise e automação fragmentada.
Instalação
npx convez-ai varunk130/ai-marketing-os1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Automação de funções de marketing (SEO, growth, outbound, sales pipeline) sem contratar especialistas
- Geração de artefatos operacionais (lead lists, A/B test plans, ROI reports, audit scorecards)
- Redução de custo de ferramentas (eliminando stacks de $20–50K/mês)
Veredito

Projeto ambicioso que demonstra como arquitetar um time IA multi-agente com ownership claro e skills reutilizáveis. Bem-executado em TypeScript/Next.js, mas é principalmente um showcase conceitual com dados sintéticos — produção real exigiria integração com APIs externas e treinamento de modelos para métricas do mundo real.
README
A tradução para o português deste README ainda está na fila.
AI Marketing OS
What this is
ai-marketing-os is a cinematic, end-to-end showcase site for the AI Marketing OS — a thesis that modern marketing teams should be operated as an orchestrated team of AI agents, not a stack of seven SaaS tools and a few scarce specialists.
It pairs a narrative landing experience with a fully interactive Launch mission that streams agent handoffs, console output, and typed artifacts in real time. The pre-scripted mission timeline runs entirely client-side, so the demo is reproducible and never burns API credits.
Disclaimer: this site is a demo. All scenarios, agents, missions, artifacts, metrics, and figures shown are hypothetical and use synthetic data created for demonstration purposes only.
The pitch
Modern marketing teams face four compounding problems:
- Expertise is locked in specialists — a senior CRO consultant is $5K/audit; a real growth scientist is rare.
- Tooling sprawl — Optimizely + Outreach + Clay + HubSpot + ZoomInfo + Ahrefs stacks to $20–50K/month.
- Generic AI doesn't solve it — ChatGPT can write copy, but it won't run your sample-size calculation or score your pipeline with a real model.
- Prompts die in chat windows — experiments are private threads with no compounding system.
The AI Marketing OS reframes this as a 4-layer architecture: skills → agents → orchestration → artifacts.
The architecture
| Layer | Count | Role |
|---|---|---|
| Skills | 12 | Markdown methodologies (Bayesian A/B math, MEDDPICC, CRO audits, cohort LTV, …) |
| Agents | 9 | Named specialists with clear ownership that load one or more skills |
| Orchestration | — | Agents hand off typed artifacts along a graph; the data flow is the program |
| Artifacts | ∞ | Real deliverables — CSVs, plans, scorecards, ROI reports — that ship |
The 9 agents
| Agent | Role | Call sign |
|---|---|---|
| Atlas | SEO | Mapper of demand |
| Echo | Content | Voice of the brand |
| Nova | Growth | Bayesian and bold |
| Lens | Conversion | Funnel surgeon |
| Forge | Outbound | Sequence architect |
| Scout | Sales Pipeline | Signal in the noise |
| Anvil | Sales Playbook | Deal carpenter |
| Oracle | Revenue Intelligence | Reads the losses |
| Ledger | Finance | Unit economics conscience |
The 12 skills
Grouped by category:
- Growth: growth-engine, conversion-ops, creative-ops, content-ops, seo-ops, podcast-ops
- Sales: outbound-engine, sales-pipeline, sales-playbook, revenue-intelligence
- Ops: finance-ops, team-ops
The upstream methodology library lives at ai-marketing-claude-skills.
Pages
| Route | Purpose |
|---|---|
/ | Overview landing: hero, problem, solution, extent, why-now, how-it-works, skills, missions, CTA |
/how-it-works | Architecture deep-dive: layer-by-layer walkthrough with code panels and contrast tables |
/skills | The 12 skills cataloged by category, each linking to its upstream README |
/missions | Mission scenarios (Launch a Product, Rescue a Stuck Pipeline, A Week at the AI Marketing Org) |
/missions/launch | Fully interactive Launch mission with live agent graph + console + artifact feed |
/about | Why this exists + related portfolio |
Quickstart
git clone https://github.com/varunk130/ai-marketing-os.git
cd ai-marketing-os
npm install
npm run dev
Then open http://localhost:3000.
Scripts
| Script | What it does |
|---|---|
npm run dev | Start the Next.js dev server with Turbopack |
npm run build | Production build |
npm start | Serve the production build |
npm run lint | Run ESLint |
Project structure
.
├── app/ # Next.js app router
│ ├── about/ # About page
│ ├── how-it-works/ # Architecture deep-dive
│ ├── missions/ # Mission scenarios
│ │ └── launch/ # Interactive Launch mission
│ ├── skills/ # Skills catalog
│ ├── globals.css # Design tokens + Tailwind v4 theme
│ ├── layout.tsx # Root layout (header + footer)
│ └── page.tsx # Home (overview)
├── components/
│ ├── mission/ # Mission runtime (console, graph, widgets, …)
│ ├── overview/ # Landing-page sections
│ ├── site/ # Header + footer chrome
│ └── ui/ # Reusable primitives
├── data/
│ ├── agents.ts # 9 specialist agents
│ ├── skills.ts # 12 skills
│ ├── missions.ts # Mission summaries
│ └── missions/launch-timeline.ts
├── lib/
│ ├── hooks/use-typewriter.ts
│ ├── mission-store.ts # Zustand store for mission playback
│ └── utils.ts
└── public/
└── hero.svg # README + OG hero image
Tech stack
- Framework: Next.js 16 (App Router, Turbopack)
- UI: React 19 + Tailwind CSS v4
- Motion: Framer Motion 12
- Graph: @xyflow/react for the live agent graph
- State: Zustand for mission playback
- Icons: lucide-react
- Fonts: Geist Sans + Geist Mono
- Language: TypeScript 5
Design system
The site uses a custom dark-first design system defined in app/globals.css:
- A 5-step surface ladder for clear visual hierarchy
- A 4-step foreground ladder for accessible text contrast
- A vibrant brand palette: violet, cyan, pink, amber, emerald, indigo, rose
- Animation utilities for drift, soft pulse, and shimmer
Contributing
This is a personal portfolio project. The main branch is protected:
- The repository owner can push directly when needed
- All other contributions must come through a pull request and require owner approval (enforced via
CODEOWNERSand branch protection)
If you spot a bug or have a suggestion, open an issue — happy to chat.
Related projects
This site is part of a broader portfolio of AI-builder repositories:
ai-marketing-claude-skills— the 12-skill source library this site visualizesai-pm-agents-suite— a 6-agent PM pipeline + 3 standalone PM agentsai-gtm-skill-library— 37 opinionated GTM skillsai-strategic-partnerships-skills— 12 skills for partnership deals and agentic trust termsai-ux-skill-library— 13 frameworks for UX of AI productsai-workflow-playbooks— 21 playbooks + 10 skills + 4 guardiansclaude-code-skills— 42 production-grade skillsai-legal-agents-skills-os— one master legal agent, 9 specialist skillsai-growth-os— a compounding growth loop: find → design → ship → learnai-revops— 9 agents take a company into a new market end to end
Author
Built by Varun Kulkarni — AI Builder at Microsoft, Forbes Tech Council member, and author of a portfolio of AI agent and skill libraries spanning product, GTM, UX, and decision-making.
License
Released under the MIT License.