How adityaprakash-work/realtor-agent-main works
Simplified High-Level Logic & Process Flow (for non-technical stakeholders)
User Interface
• Streamlit frontend or any API client where a user enters their search.
Prototype Server
• FastAPI service that receives requests and coordinates work.
Orchestrator & Agents
• Internal “brains” that decide which external service to call (e.g., Zillow search).
External Real-Estate APIs
• Third-party data sources (Unofficial Zillow API via RapidAPI).
AI Processing
• Optional enrichment via OpenAI / GoogleAI models.
Response Formatter
• Final step that packages results for display.
User → UI
User enters property search criteria (location, price range, etc.).
UI → Server
The UI sends a request to the Prototype Server’s chat/search endpoint.
Server → Orchestrator/Agents
The server’s orchestrator hands off the request to the appropriate agent (e.g., Zillow agent).
Agent → External API
The agent calls the real-estate API to fetch listings.
External API → Agent
Listings data returns to the agent.
Agent → AI Processing (if enabled)
The server may send raw listings through an AI model to refine recommendations or add commentary.
AI → Server
Enriched data comes back.
Server → Response Formatter
The server packages the data into a neat response format.
Server → UI
The formatted results are sent back to the UI and displayed to the user.
• Blue for UI
• Green for Server & Orchestrator
• Orange for External APIs
• Purple for AI Processing
This high-level flow captures the core steps without technical jargon, making it easy for business stakeholders to understand how user requests transform into real-estate search results.