IntelSync: Multi-Agent Market Intelligence

• 

  Diagram of the technology architecture.

• 

  Dashboard result generated by the pipeline, part 1.

• 

  Dashboard result generated by the pipeline, part 2.

Inspiration

IntelSync was born from the need for real-time market intelligence: scattered news sources, manual analysis, and slow decision cycles inspired a fully automated, multi-agent solution.

What it does

Automatically scrapes configured web sources, loads raw data into BigQuery, enriches it with sentiment and entity analysis via Cloud NLP, and presents insights in an interactive dashboard.

How we built it

• Agent Development Kit (ADK): orchestrates WebScraperAgent, BigQueryLoaderAgent, and InsightGeneratorAgent
  
• Google Cloud BigQuery: scalable data storage
  
• Cloud Natural Language API: sentiment and entity extraction
  
• Streamlit: real-time, exportable dashboard
  

Challenges we ran into

• Coordinating asynchronous agent workflows
  
• Handling API rate limits and retries for web scraping
  
• Managing GCP credentials and permissions across services
  

Accomplishments that we’re proud of

• A robust end-to-end multi-agent pipeline using ADK
  
• Open-source contribution to the ADK repo
  
• Fully deployed, zero-ops Streamlit dashboard with live filtering and download features
  

What we learned

• Multi-agent orchestration unlocks powerful automation patterns
  
• Deep integration with Google Cloud services simplifies complex NLP workloads
  
• A polished UX is critical for rapid insight adoption
  

What’s next for IntelSync: Multi-Agent Market Intelligence

• Deploy to Cloud Run for zero-ops hosting and auto-scaling
  
• Automate daily refresh with Cloud Scheduler & Functions
  
• Expand agent library to support custom data sources and advanced models
  

Built With

• agent-development-kit
• altair
• git
• github
• google-bigquery
• google-cloud
• natural-language-processing
• python
• streamlit