TrendTracker - A Multi-Agent System for Analyzing and Generating New Ideas

This is a submission for the Agent.ai Challenge: Assembly of Agents (See Details)

What I Built

TrendTracker is a two-agent system designed to identify trending ideas from challenge submissions and generate new concepts or improvements based on those trends.

1. ScraperAgent:

   • Gathers data (titles, descriptions, reaction counts) from DEV.to challenge submissions.
   • Stores the scraped results for further analysis.

2. TrendTracker Agent:

   • Invokes the ScraperAgent to collect the latest data.
   • Analyzes the scraped results, ranking submissions by popularity.
   • Generates creative ideas or improvements inspired by the top-performing submissions.

I built this system to streamline the discovery of trending topics and to spark innovation for anyone looking to create an AI-driven solution. Imagine a scenario where you want to see what’s resonating with users, then quickly brainstorm new features, side projects, or product enhancements. TrendTracker makes that process seamless.

Demo

https://agent.ai/agent/trendtracker

Below is a quick walkthrough of how the system works:

• Step 1: ScraperAgent scrapes the DEV.to challenge page for new submissions.
• Step 2: TrendTracker calls the scraper’s output, runs an LLM prompt to analyze the data, and generates suggestions.
• Step 3: You receive a ranked list of top ideas, along with improvement proposals or entirely new agent concepts.

Agent.ai Experience

Building with the Agent.ai Builder was straightforward and fun. The “Assembly of Agents” concept allowed me to chain multiple agents together easily. Here’s what stood out:

• Delightful Moments:

  • The drag-and-drop SmartFlow interface for chaining the scraper and idea-generator.
  • Being able to store intermediate results and pass them along to the next agent.
  • Quickly iterating on prompts without leaving the browser interface.

• Challenges:

  • Making sure the scraper returns clean, structured data so the TrendTracker agent can interpret it accurately.
  • Refining the LLM prompt to focus on creative suggestions rather than just summarizing.

Overall, the combination of multi-agent orchestration and prompt-driven AI made it easy to build a powerful system in a short amount of time.