Content Marketing Agent

AI-powered content studio that generates, validates, schedules, and analyzes multi-platform marketing content end-to-end.

🗺️ Architecture diagram

For GitHub README (place the image file at the repo root or update the path accordingly):

---

---

🚀 Quick pitch

• Create brand-safe content in seconds
• Approve with automatic compliance checks
• Publish everywhere and track results in real time

🎯 Problem

Creating consistent, on-brand content across multiple platforms is slow and error-prone. Teams bounce between tools for ideation, creation, approval, scheduling, and analytics.

💡 Our solution

A single AI-driven content studio that:

• Generates platform-optimized content using Amazon Bedrock Nova
• Enforces brand guidelines automatically
• Surfaces trend-backed ideas
• Publishes to Twitter, LinkedIn, Facebook, and Instagram
• Learns from performance to improve future content

---

✨ Key features

• AI content generation with brand context and tone control
• Trend-backed content suggestions from Twitter, Google Trends, and Reddit
• One-click publish and cross-platform scheduling
• Real-time analytics and engagement prediction
• Role-based access with secure key management

---

🧱 Architecture overview

• Pattern: Microservices‑oriented monolith, 3‑tier (Presentation, Business Logic, Data)
• Frontend: React 18 + TypeScript + Vite + Tailwind
• Backend: FastAPI (Python) + Uvicorn, Async SQLAlchemy, Pydantic
• Storage: SQLite (dev), PostgreSQL (prod)
• Cache/queues: Redis
• Cloud: AWS (Bedrock Nova, S3, RDS, ElastiCache)
• Deploy: Docker, GitHub Actions, Terraform/CloudFormation

Core services

• ContentGenerator (Bedrock integration)
• BrandValidator (Compliance engine)
• TrendMonitor + UnifiedTrendsAggregator
• PerformanceLearner (ML insights)
• Config and Credential managers

External integrations

• Amazon Bedrock: amazon.nova-pro-v1:0 for content generation
• Social APIs: Twitter v2, LinkedIn, Facebook Graph, Instagram Basic Display

---

📦 Tech stack

• Frontend: React 18, React Router, React Query, Tailwind, Lucide, Axios
• Backend: FastAPI, Uvicorn, SQLAlchemy 2.0 (async), Alembic, Pydantic v2
• Infra: Docker, Redis, PostgreSQL, AWS (S3, Bedrock, RDS), GitHub Actions
• Observability: Sentry, CloudWatch, structured JSON logs

---

🛠️ Getting started (local)

1. Prereqs: Node 20+, Python 3.11+, Docker, Redis
2. Clone and configure

cp .env.example .env               # fill values below
cp frontend/.env.example frontend/.env

1. Install and run

# backend
python -m venv .venv && source .venv/bin/activate
pip install -r backend/requirements.txt
alembic upgrade head
uvicorn backend.main:app --reload --port 8001

# frontend
cd frontend
npm i
npm run dev -- --port 3000

🔐 Required environment variables

Backend

BEDROCK_REGION=us-east-1
AWS_ACCESS_KEY_ID=...
AWS_SECRET_ACCESS_KEY=...
JWT_SECRET=change-me
DATABASE_URL=postgresql+asyncpg://user:[pass@localhost:5432](mailto:pass@localhost:5432)/content
REDIS_URL=redis://[localhost:6379](http://localhost:6379)
S3_BUCKET=content-agent-assets-prod
CORS_ORIGINS=http://localhost:3000

Social integrations

TWITTER_API_KEY=...
TWITTER_API_SECRET=...
LINKEDIN_CLIENT_ID=...
LINKEDIN_CLIENT_SECRET=...
FACEBOOK_APP_ID=...
FACEBOOK_APP_SECRET=...
INSTAGRAM_CLIENT_ID=...
INSTAGRAM_CLIENT_SECRET=...

---

▶️ Run with Docker

docker compose up --build

• Frontend: http://localhost:3000
• Backend: http://localhost:8001
• Health check: GET /health

---

📚 API at a glance

Base: /api

Content

• POST /content/generate — AI content generation
• GET /content/{id} — retrieve
• PUT /content/{id} — update
• POST /content/{id}/publish/{platform} — publish
• POST /content/{id}/schedule — schedule

Suggestions

• GET /content-suggestions — list
• POST /content-suggestions/generate-fresh — new AI suggestions
• POST /content-suggestions/{id}/review — approve or reject
• POST /content-suggestions/{id}/publish — publish suggestion
• POST /content-suggestions/{id}/schedule — schedule suggestion

Brands

• GET /brands — list
• POST /brands — create
• GET /brands/status/active — active brand
• POST /brands/{id}/actions/set-active — set active
• PUT /brands/{id} — update

Config

• GET /config/status — system status
• GET /config/keys — masked keys
• PUT /config/keys — update keys
• POST /config/test-connection/{service} — verify integrations
• POST /config/reload — reload configuration

Auth

• /auth — JWT-based authentication

---

🗄️ Data model (high level)

Contents: id, type, title, body, platform, brand_id, status, performance_metrics, timestamps

BrandGuidelines: id, brand_name, voice_tone, key_messages, prohibited_words, traits, demographics, timestamps

Campaigns: id, name, goals, budget_allocation, status, metrics, timestamps

Users: id, email, role, password_hash, flags, timestamps

---

🔒 Security

• JWT with access 24h and refresh 30d
• Encrypted API keys at rest, HTTPS in transit
• Bcrypt password hashing, input validation, rate limiting
• CORS, SQLi and XSS protections

---

📈 Monitoring

• /health aggregated checks for DB, cache, external APIs, AI
• Metrics: latency, error rates, AI generation metrics, API quotas
• Logging: structured JSON, Sentry for errors

---

✅ What I built during the hackathon

• End-to-end flow: generate → validate → schedule/publish → track
• Bedrock Nova integration with prompt templating
• Brand compliance engine with rules and prohibited lexicon
• Trend ingestion from Twitter, Google Trends, Reddit with unified ranking
• Basic analytics with engagement prediction

🗺️ What’s next

• Team collaboration and review workflows
• Canva-style editor for rich media
• Auto A/B testing with multi-armed bandits
• More platforms and webhooks

---

📄 License

MIT (placeholder)

👥 Team

• Pranav

---

Appendix: Original architecture notes

Expand to view the detailed architecture captured during exploration

Content Marketing Agent - Architecture Overview

🏗️ System Architecture

High-Level Architecture Pattern

Pattern: Microservices-oriented monolith with modular components

Architecture Style: 3-tier architecture (Presentation, Business Logic, Data)

Communication: RESTful APIs with real-time capabilities

Deployment: Containerized with Docker, cloud-ready

[The detailed architecture, routes, models, security, monitoring, deployment notes have been preserved in the source page history.]