MaatramMark

COMPLETE WORKFLOW OF MAATRAM MARK
HOME PAGE
UNDERSTANDING MAATRAM MARK
ONBOARDING PAGE
DASHBOARD PAGE
VISUAL CONTENT GENERATION
FEEDBACK PAGE
REELS STORY BOARD PAGE

Inspiration

Small businesses and local brands often struggle to create engaging social media content consistently. Managing posters, captions, hashtags, and reel ideas every day requires time, creativity, and marketing knowledge. We wanted to build a system that simplifies this process using artificial intelligence. The idea behind MaatramMark was to create an AI-powered marketing assistant that not only generates content automatically but also learns from audience engagement and improves future recommendations over time.

The name “MaatramMark” comes from the Tamil word “Maatram,” which means change or transformation. Combined with the word Mark, representing marketing and brand identity, the name reflects the core vision of the project — transforming digital marketing using AI.

What it does

MaatramMark is an AI-powered social media marketing assistant designed for businesses and creators. The platform generates daily marketing content such as posters, Instagram reel ideas, captions, hashtags, and branding suggestions using multiple AI agents.

The system includes:

AI-generated poster prompts
Instagram reel ideas and scripts
Captions and hashtags
Branding and strategy suggestions
Daily content recommendations
Feedback-driven learning system

Users can customize AI-generated prompts before generating images. After posting content on social media, they can submit engagement metrics such as likes, comments, shares, and reach. The AI analyzes this feedback and improves future content recommendations automatically.

How we built it

We built the frontend using React to create a modern dashboard-style interface. The backend was developed using FastAPI, where we implemented a modular AI-agent architecture.

We used:

OpenRouter for LLM-powered AI agents
Clipdrop API for AI-powered poster generation
Supabase for database management and storage
Supabase Storage Buckets for storing posters and logos

The system consists of multiple AI agents including:

Strategy Agent
Branding Agent
Research Agent
Reels Agent
Feedback Learning Agent

The /update-today endpoint acts as the central orchestrator that combines business information, branding data, trends, and feedback memory to generate personalized daily marketing content.

Challenges we ran into

One of the biggest challenges was designing a clean multi-agent AI workflow without making the backend overly complicated. We also faced difficulties while integrating image generation services because many free APIs produced inconsistent or low-quality outputs.

Other major challenges included:

Managing Supabase Row Level Security (RLS) policies
Handling storage bucket permissions
Integrating AI-generated images with cloud storage
Prompt engineering for consistent AI responses
Designing the feedback-learning loop effectively
Deployment and API integration issues

Creating a system where the AI genuinely learns from user engagement instead of simply generating static content was one of the most difficult and important parts of the project.

Accomplishments that we're proud of

We are proud of building a complete AI-powered marketing workflow that combines content generation, personalization, and feedback learning into one unified platform.

Some key accomplishments include:

Creating a modular AI-agent architecture
Implementing personalized daily content generation
Building a feedback-driven learning system
Generating both poster and reel content dynamically
Integrating cloud storage for AI-generated assets
Designing a scalable FastAPI backend structure

We are especially proud that the system continuously improves its recommendations based on user engagement data.

What we learned

Through this project, we learned:

Full-stack AI application development
FastAPI backend architecture
Prompt engineering techniques
AI orchestration using multiple agents
Supabase database and storage management
API integration and deployment workflows
Feedback-driven AI personalization

Most importantly, we learned how AI systems can move beyond simple generation and become adaptive systems that improve continuously using real-world data.