Agentic AI POC Architecture

System Overview

Single FastAPI process exposing many agentic capabilities via modular routes and services. Shared NLP utilities support keyword extraction and semantic scoring. Optional browser automation for scraping/LinkedIn.

Architecture Diagram

graph TB
    FastApiEntry["app.py<br/>FastAPI entry"]
    RouteModules["routes<br/>rfp, scraper, ted, linkedin, n8n, summarize"]
    ServiceLayer["services<br/>business logic"]
    UtilsHelpers["utils<br/>NLP + extraction helpers"]
    SentenceModel["sentence-transformers model<br/>all-MiniLM-L6-v2"]

    FastApiEntry -->|"registers HTTP routes"| RouteModules
    RouteModules -->|"calls service layer"| ServiceLayer
    ServiceLayer -->|"uses NLP / scrape helpers"| UtilsHelpers
    UtilsHelpers -->|"runs embeddings / similarity"| SentenceModel

How to read this diagram

Arrows mean flow / dependency / "talks to", not button clicks. Each arrow label says what happens on that step.

Route modules

• rfp_routes.py
• scraper_routes.py / scraper_alias_routes.py
• ted_routes.py / ted_alias_routes.py
• linkedin_routes.py / linkedin_alias_routes.py
• link_summarize_routes.py
• n8n_routes.py
• frontend_routes.py

Service modules

• comparison_service.py
• scraper_service.py
• ted_scraping_service.py, ted_semantic_service.py, ted_summarization_service.py, ted_email_service.py
• link_summarization_service.py
• url_service.py

Technology Stack

Layer	Technology
API	FastAPI
Embeddings	sentence-transformers
Scraping	Playwright / Selenium
Automation glue	n8n helpers

Next Steps

• API Reference
• Getting Started
• Overview