CreatorCopilot

AI Agent MVP 7 min read

Product Vision

CreatorCopilot is an AI-powered content strategy agent that helps content creators and small businesses generate high-performing social media campaigns. This portfolio project demonstrates advanced AI agent architecture using OpenAI's function calling, integration with GPT-4o API, REST API orchestration, and modern cloud application design.

Key deliverables:

AI agent using OpenAI function calling

GPT-4o with autonomous tool use capabilities

REST API integration (SerpAPI for trend research)

Full-stack web application (React + FastAPI)

Core User Problem

Content creators and small businesses spend significant time on manual tasks:

Brainstorming content ideas aligned with current trends
Researching what's performing well in their niche
Writing compelling scripts and hooks
Creating captions and selecting relevant hashtags
Planning video shoots and shot lists

CreatorCopilot reduces this manual effort by acting as an AI content strategist that autonomously researches trends, analyses business context, and generates complete campaign strategies.

MVP Scope

The MVP is a working web application where users input business details and receive AI-generated campaign strategies.

Demo

Note: Please update the API key field with your API key in a .env folder before running the application.

Input Fields:

Business name and industry/category
Location and target audience
Social media platform (TikTok, Instagram, YouTube)
Campaign objective (awareness, engagement, conversions)
Brand tone (authentic, casual, professional, playful)

Example Input:

Business: Japanese restaurant

Location: Sydney

Platform: TikTok

Objective: Increase restaurant awareness

Audience: Sydney food lovers

Tone: Authentic, casual, engaging

AI-Generated Output:

Campaign concept and content angle
Attention-grabbing video hook
30-60 second script optimised for platform
Detailed shot list for video production
Engaging caption with storytelling elements
Relevant hashtags based on trends and platform
Clear call-to-action
Content optimisation recommendations

Why OpenAI GPT-4o?

This project uses OpenAI's GPT-4o with function calling for several strategic reasons:

Advanced Function Calling: GPT-4o's function calling capabilities enable autonomous decision-making, determining when and how to call external APIs
Global Availability: OpenAI's API has broad regional support, making it accessible for developers worldwide including Australia
Production-Ready: OpenAI's mature API and extensive documentation provide reliable patterns for building scalable AI agents
Content Quality: GPT-4o demonstrates exceptional performance in creative content generation, storytelling, and marketing copy
Cost Efficiency: Competitive pricing with flexible usage tiers suitable for both prototyping and production deployment

AI Agent Architecture

CreatorCopilot implements a true AI agent workflow, where the agent autonomously determines which actions to take and orchestrates multiple tools to complete the task using OpenAI's function calling.

Agent Workflow:

OpenAI Function Calling:

The agent uses OpenAI's function calling capability to intelligently decide when and how to use available tools. The agent architecture provides a structured framework for orchestrating multi-step workflows, enabling autonomous decision-making rather than rigid, predefined workflows.

Tool 1: Trend Research

Purpose: Retrieve trending topics, keywords, and content patterns related to the user's business and industry.

Implementation: Integration with SerpAPI (or similar search/trends API) to gather real-time trend data.

Example Flow:

Input: "Japanese restaurant Sydney"

↓

API Call: SerpAPI search query

↓

Output: Trending keywords, popular searches, related topics

↓

Agent Context: Uses trend data to inform content strategy

This tool ensures that generated campaigns are grounded in current trends and real search behaviour, not just generic templates.

Tool 2: Content Strategy Generator

The AI agent analyses multiple inputs to create a comprehensive content strategy:

Business type and unique value proposition
Target audience demographics and psychographics
Platform-specific algorithm preferences and best practices
Trending topics from research tool
Campaign marketing objective

Structured JSON Response:

{

  "campaign_name": "Behind the Scenes: Authentic Ramen",

  "content_angle": "Show traditional ramen-making process",

  "hook": "POV: You're about to discover why...",

  "script": "30-60 second narrative script",

  "shot_list": [

    "Close-up: Chef hand-pulling noodles",

    "Wide shot: Kitchen atmosphere",

    "Close-up: Broth simmering"

  ],

  "caption": "Engaging caption with storytelling",

  "hashtags": ["#SydneyEats", "#RamenLovers"],

  "cta": "Visit us at [location]",

  "recommendations": [

    "Post during peak engagement hours",

    "Use trending audio"

  ]

}

Tech Stack

Frontend: React + Vite + Tailwind CSS

Clean, modern UI with responsive design
Loading states and error handling
Three core pages: Landing, Generator, Results
Form validation and user feedback

Backend: Python FastAPI

RESTful API endpoints
POST /generate-campaign as the Main agent workflow
GET /health as the Health check endpoint
Async processing for API calls
Error handling and logging

AI Implementation: OpenAI GPT-4o

OpenAI function calling for workflow orchestration
Autonomous tool selection and execution
System prompts optimised for content strategy
128K token context window for comprehensive analysis
Structured outputs with JSON formatting

External APIs:

OpenAI API for LLM inference (GPT-4o)
SerpAPI for trend research and search data

AI Agent System Prompt

System Role:

"You are an expert social media strategist specialising in

short-form content marketing. You understand audience psychology,

storytelling hooks, retention tactics, and platform algorithms."

Agent Reasoning Process:

1. Understand campaign objective

2. Research relevant trends (call trend tool)

3. Analyse business context and audience

4. Develop platform-optimised content strategy

5. Generate structured campaign deliverables

Development Approach

This MVP prioritises core functionality over unnecessary complexity:

Focus Areas:

Working AI agent with real tool integration
Clean, scalable architecture
Excellent user experience
Clear demonstration of technical capabilities

Not in Scope:

User authentication (not required for demo)
Payment processing (not relevant for MVP)
Complex databases (in-memory state sufficient)
Microservices architecture (unnecessary complexity)

Deployment Strategy

Frontend Deployment:

Platform: Vercel
Automatic deployments from Git
Environment variables via Vercel dashboard
Global CDN for optimal performance

Backend Deployment:

Platform: Render or AWS (containerised)
Environment variable configuration for API keys
Health check endpoint monitoring
Auto-scaling based on traffic

Required Environment Variables:

OPENAI_API_KEY=your_openai_api_key

SERPAPI_KEY=your_serpapi_key

FRONTEND_URL=https://your-frontend.vercel.app

ENVIRONMENT=production

Next Steps & Extensions

While this MVP focuses on core functionality, there are future iterations that can be embedded into the agent:

Technical Extensions:

Multi-platform campaign generation (TikTok, Instagram, YouTube simultaneously)
Image generation integration (DALL-E for thumbnail concepts)
Campaign performance prediction using ML models
A/B testing recommendations based on historical data

Product Extensions:

Campaign history and iteration capabilities
Export to PDF or presentation format
Analytics integration for performance tracking

Conclusion

CreatorCopilot demonstrates how AI agents can transform creative workflows by combining intelligent decision-making, external data integrations, and user-centric product design. By developing an AI-powered content strategist that assists with ideation, campaign planning, and content optimisation, this project showcases the potential of agentic AI to solve real-world business challenges.