Control in mobile view
Live log in mobile view
How to use this?
Main Dashboard

NewsifyX.ai

Inspiration

In an era of 24/7 news cycles, information overload makes it impossible for individuals to keep up with every major global event. Traditional news aggregators are often passive, requiring users to manually check for updates. I wanted to build a "Digital Journalist"—an autonomous entity that doesn't just collect links, but understands the weight of global news and broadcasts it with the speed of a professional breaking-news desk.

By leveraging Groq’s ultra-low latency inference, I realized I could create an agentic system that thinks and acts in near real-time, making autonomous social media management faster and more intelligent than ever before.

What it does

NewsifyX.ai is an autonomous AI agent that manages a 24/7 news broadcast on X (Twitter).

Scrapes & Curates: Monitors global RSS feeds and news sources continuously.
Agentic Planning: A Planner Agent evaluates news for relevance and impact, deciding what is "tweet-worthy."
Intelligent Writing: Using Llama 3.3 70B, it crafts punchy, context-aware tweets with relevant hashtags and summaries.
Autonomous Broadcasting: Automatically posts to X via API without human intervention.
Mission Control: A real-time React Dashboard allows users to monitor the agent's "thought process," view logs in Firebase, and manually trigger cycles if needed.

How we built it

The architecture is designed for speed and reliability:

The "Brain": Powered by Groq API (Llama 3.3 70B). Groq's LPU™ (Language Processing Unit) architecture was essential for the agentic workflow, as the Planner and Controller need to "communicate" back and forth rapidly.
Agentic Framework: I implemented a Planner-Controller architecture. The Planner outputs a structured JSON "Mission Plan," which the Controller then executes using various tools.
Backend: Built with FastAPI for high-performance asynchronous task handling.
Real-time Sync: Firebase Firestore acts as the central nervous system, syncing the agent's status and logs to the React 19 frontend instantly.

Challenges we ran into

API Rate Limiting: X (Twitter) has strict rate limits. I had to build a "Self-Healing" mechanism with exponential backoff to ensure the agent doesn't get blocked.
Structured Output: Getting LLMs to consistently output valid JSON for the Planner Agent can be tricky. I solved this by using strict Pydantic models and specific system prompting to ensure the pipeline never breaks.
Context Preservation: Distilling long news articles into 280-character tweets while maintaining accuracy required fine-tuning the prompt instructions to avoid "hallucinations."

Accomplishments that we're proud of

Inference Speed: We achieved sub-second reasoning times. If is the time taken to generate a plan and a tweet, our optimization with Groq yielded:

This is significantly faster than standard cloud-hosted LLMs.

Full Autonomy: The system can run for days without a single manual refresh, successfully handling news spikes and network fluctuations.
Dashboard Design: Created a professional-grade monitoring UI that makes an "invisible" AI process completely transparent to the user.

What we learned

Agents over Chatbots: I learned that true AI utility comes from Agency—the ability for a model to use tools and make decisions—rather than just answering questions.
Latency is the Key to Agency: High-speed inference (like Groq) is not just a luxury; it's a requirement for multi-step agentic workflows where latency accumulates at every step.
State Management: Learned how to effectively use Firebase for real-time status tracking in a distributed system.

What's next for NewsifyX.ai

Multimodal Updates: Integrating image generation models to create a custom "news graphic" for every tweet.
Sentiment-Driven Posting: Adjusting the "voice" of the tweet based on news sentiment (e.g., serious for hard news, energetic for tech breakthroughs).
Multi-Platform: Expanding the agent's reach to LinkedIn and Threads to create a cross-platform news empire.

This video provides a deep dive into building autonomous agents using Groq and Llama, which mirrors the core technology stack used in NewsifyX.ai.