Project Story
Inspiration:
As the PR for my college’s student club, I constantly face a challenge: we host events frequently, often with little notice. This leaves me scrambling to create quality social media content on time, resulting in posts that don’t fully capture our events.
So I decided to join the Redacted hackathon; I saw it as the perfect opportunity to solve this problem. But I had just 10 days to make it happen.
What We’ve Built:
With time ticking, I split the project into three core parts—AI, frontend/backend, and deployment—each working together to make the platform come alive:
AI-Driven Content Creation
The journey started with developing an AI model trained on MASA’s Twitter data from five active accounts focused on crypto, Web3, and blockchain.
I spent nearly five days ensuring it was finely tuned to capture trends, sentiment, and popular discussions in these fields.
Using Crew AI Agentic Workflow and OpenAI for text generation, the model crafts high-quality, tailored content across platforms, creating post-ready material for Twitter, Instagram, LinkedIn, and YouTube.
Custom Image Generation with Flux Schnell
For visual content, the AI creates image descriptions that are transformed into prompts, which are then sent to the Replicate Flux Schnell API.
This integration generates unique images for each social media post, enhancing engagement and resonating with the crypto, Web3, and blockchain themes we wanted to highlight.
Frontend Design & Backend with Flask
I had a clear vision of the platform I wanted, but with no experience in building a website, especially with Next.js and backend development, I faced a learning curve.
Using Cursor and Claude AI as supports, I designed an accessible, user-friendly interface, and for the backend, I learnt Flask to manage the server-side logic. Together, they made for a streamlined platform, where users can easily generate and manage high-quality content with minimal input.
Deployment
Deployment was the final hurdle—and probably the toughest. With just two days left and almost no experience in deployment, I had two options: push through and risk errors, or present my project story.
I chose to prioritize stability, even if it meant presenting the project without a full deployment.
Going forward, I plan to deploy the platform completely, ensuring it’s robust, accessible, and ready for real-world use.
Potential Impact on NEAR Ecosystem
This project has the potential to contribute significantly to the NEAR ecosystem. The platform could serve as a content engine for MASA and others within NEAR, helping them produce timely, high-quality social media content. This would allow Web3 projects to maintain an authentic voice, reach a broader audience, and engage more effectively within a decentralized, community-oriented landscape.
What We Wanted to Add
Though we built a powerful content generation platform, time constraints kept us from a few additional features we had hoped to implement:
- We wanted to add real-time Twitter data integration to keep the AI model updated with the latest trends and discussions, making the generated content even more relevant.
- To further streamline the process, we envisioned a one-click posting feature that would allow users to publish the generated content directly to their social media accounts.
- Another goal was a dedicated Reels section for Instagram, where the AI could generate ready-to-upload reel videos, helping users engage with audiences through video content.
The Bigger Picture
This hackathon taught me that innovation isn't just about technical achievements - it's about solving real problems for real people.
Our platform isn't just generating content; it's helping creators and businesses maintain a consistent, professional presence across the social media landscape in the rapidly evolving world of crypto, Web3 and blockchain.
The journey from concept to creation was challenging, enlightening, and ultimately rewarding. While we may not have implemented every planned feature, we built something that works, something that matters, and something that can grow.
Built With
- ai
- amazon-web-services
- api
- claude
- crew
- cursor
- ec2
- flask
- flux
- javascript
- masa
- milvus
- next.js
- openai
- replicate
- schnell
Log in or sign up for Devpost to join the conversation.