The Core Stack
From client request to final structured intelligence.
Frontend (Next.js)
Firebase Hosting
API Layer (Functions)
Research Orchestrator
Search + Scraping Layer
AI Processing (Groq)
Verification Engine
Citation Engine
Final Response
The AI Research Pipeline
The step-by-step autonomous lifecycle of a single query.
01
User Query
Initial prompt ingestion.
02
Intent Analysis
Decoding user goals (informational, comparative, analytical).
03
Query Expansion
Generating 5-10 sub-queries to ensure comprehensive search coverage.
04
Source Discovery
Pinging Brave, Tavily, and SerpAPI for highest-ranked URLs.
05
Scraping (Live Web)
Playwright agents bypass blockers and download raw page DOMs.
06
Extraction
Cleaning HTML into pure text, preserving tables and stats.
07
Verification
Cross-referencing claims across multiple scraped sources.
08
Reasoning
DeepSeek/Llama models synthesize data into logical arguments.
09
Citation Generation
Mapping specific sentences to their exact URL source.
10
Final Report
Delivery of structured Markdown with integrated citations to UI.