1
Step 1: Data Hunters Contribute Content
The process begins with our global network of Data Hunters. Using a browser
extension, they capture relevant social media posts, news articles, and other
online content as they browse. This human-in-the-loop approach ensures that we
are sourcing content that is genuinely interesting and significant.2
Step 2: Verification and Filtering
Once submitted, the raw data enters our distributed infrastructure for validation.
Automated checks and AI classifiers work to filter out spam, duplicates, and
irrelevant content. This ensures that only high-quality, timely information proceeds
to the next stage.3
Step 3: Clustering and Analysis
The verified posts are then transformed into vector embeddings. An unsupervised
clustering algorithm groups semantically related posts into “discussion clusters,”
representing distinct narrative themes. We also apply a time-decayed engagement
score to prioritize the most active and relevant conversations.4
Step 4: API Delivery
The final, structured context is made available through our developer-friendly
API. AI developers can easily integrate this real-time data feed into their applications
using our RESTful endpoints or the Model Context Protocol (MCP) server.