Agent X

Inspiration

Trolling-as-a-Service: There are so many annoying keyboard warrior on the internet that spread misinformation or just negativity in general. If only I have an army of troller to feed them their own medicine.

Giving-a-Fuck-as-a-Service: I had a X account for almost 9 years and only got 1 follower. The only person that ever replied to my post is, well myself. This is so sad, if nobody gonna comment on me, I will build it for myself.

What it does

AgentX revolutionizes Twitter engagement by leveraging AI to generate authentically human-sounding responses across the emotional spectrum. Our platform offers two primary functions:

Trolling-as-a-Service: Creates witty, provocative replies to any tweet with customizable spiciness
Bootlicking-as-a-Service: Generates excessively flattering comments to profiles' latest tweets

The system supports both one-off responses and scheduled campaigns, allowing users to queue up to 10 successive replies with 16-minute intervals to avoid Twitter rate limits. All activities are tracked through an intuitive dashboard with real-time logging capabilities.

How we built it

We constructed AgentX using a modern, full-stack approach:

Backend: Node.js with Express powers our RESTful API, handling Twitter integration via Twitter API v2 AI Integration:
Dual AI provider system with Azure OpenAI as primary and Google Gemini as fallback
Frontend: Pure HTML/CSS/JavaScript featuring fullPage.js for seamless navigation and responsive design
Authentication: Secure login system via Supabase with JWT token management
Monitoring: Azure Application Insights for comprehensive system logging
Deployment: Azure App Service with GitHub Actions for CI/CD pipeline

The architecture follows separation of concerns with modular components handling different aspects of the application, from AI prompt generation to scheduled response queuing.

Challenges we ran into

Building AgentX presented several hurdles:

Twitter API Integration: Working within Twitter's rate limits and dealing with their complex authentication requirements
Azure OpenAI Character Limits: Ensuring responses met Twitter's 280 character limit while remaining contextually relevant
Multi-provider AI Architecture: Creating a seamless fallback system between Azure OpenAI and Google Gemini. OpenAI model is too expensive.
Scheduled Tasks: Implementing reliable background processing for timed Twitter engagement
Environment Configuration: Ensuring proper environment variable management across local and cloud deployment. Getting the key from Azure wasn't an easy task.

Accomplishments that we're proud of

Creating an AI system that generates remarkably human-sounding tweets across different emotional tones
Building a complete user management system with secure authentication
Implementing a dual AI provider architecture with automatic fallback
Developing a real-time monitoring dashboard for tracking scheduled activities
Creating an intuitive UI that makes complex AI interaction accessible to non-technical users

What we learned

Prompt engineering techniques to control AI-generated content tone and character count
Azure OpenAI integration and API behavior compared to other AI services
Building robust scheduling systems for social media automation
Managing environment configurations across development and production
Implementing effective error handling for external API dependencies

What's next for Agent X

We're excited about AgentX's future development:

Content Analysis: Implementing sentiment analysis to provide insights on engagement effectiveness
Multi-platform Support: Expanding beyond Twitter to other social media platforms
Advanced Scheduling: Adding more complex scheduling patterns and campaign management
Custom Persona Creation: Allowing users to define AI personalities for different engagement types
Enterprise Features: Developing team collaboration tools for marketing agencies

Our vision is to transform AgentX into a comprehensive AI-powered social media personality management platform for individuals and businesses alike.