AI-First HCP Interaction CRM

An intelligent CRM system for Life Sciences field representatives to log and manage interactions with Healthcare Professionals (HCPs). This AI-First CRM uses a natural language AI assistant to automatically extract structured information from user prompts, eliminating manual form filling and streamlining CRM workflows.

Live Demo: https://ai-first-hcp-interaction-crm.vercel.app/

Example Prompts / Testing

You can test the AI agent using the following example prompts for each tool:

Log Interaction (log_interaction_tool)

"I met Dr Smith today at 3 PM for a meeting about Product X efficiency. John from marketing also attended. I shared brochures and the response was positive. We will follow up on 25 March."

Edit Interaction (edit_interaction_tool)

"Update the last interaction: the HCP name is actually Dr John and the sentiment should be Negative."

"For my most recent interaction, change the time to 4 PM and add that we discussed pricing as well."

Add Follow-up (add_followup_tool)

"Set the follow-up date for the last interaction to 30 March 2026."

"Please move the follow-up for my latest interaction to next Monday."

Update Materials (update_materials_tool)

"Also note that I shared the OncoBoost Phase III PDF in the last meeting."

"Update the materials shared for my last interaction to: brochures, OncoBoost Phase III PDF, dosing guidelines."

List / Get Interactions (get_interactions_tool)

"Show me the last 3 interactions and summarize their sentiments."

"List my recent interactions with Dr Smith and tell me the sentiment for each.

Overview

Users communicate with an AI assistant through natural language prompts.
The backend AI agent (LangGraph + Groq LLM) extracts structured fields:
- HCP Name, Date, Time, Topics, Materials Shared, Sentiment, Follow-up Actions, Key Outcomes
Extracted data is automatically populated in the React CRM form.
All interactions are saved in PostgreSQL via SQLAlchemy.

Tech Stack

Layer	Technology
Backend	FastAPI, Python 3.10+, LangGraph, LangChain, Groq LLM
Database	PostgreSQL, SQLAlchemy
Frontend	React 18, Redux Toolkit, Axios, CSS
Hosting	Render (backend), Vercel (frontend)

Architecture

Backend Flow

The frontend sends a POST /agent/chat request with the user’s message.
The LangGraph AI agent decides which tool to invoke:
- log_interaction – create new interaction
- edit_interaction – update specific fields
- get_interactions – fetch recent interactions
- add_followup – set or move follow-up dates
- update_materials – update shared materials
Tools read/write data in the PostgreSQL database.
The backend returns structured fields that automatically populate the frontend form.

Frontend Layout

Left Panel – AI Chat Assistant
- Accepts natural language prompts
- Sends messages to the backend AI agent
- Receives extracted structured fields
Right Panel – Interaction Form
- Auto-filled by the AI assistant
- Displays HCP interaction details (e.g., name, date, topics, materials, sentiment, follow-ups) in real-time

Project Structure & Separate READMEs

For detailed setup, architecture, and usage, see the separate README files in the repository:

Backend: backend README
Frontend: frontend README

Each README contains installation instructions, environment variable setup, API documentation (backend), and running the development server (frontend).

Key Features

AI-Driven Interaction Logging: Automatically extracts HCP name, date, topics, materials shared, sentiment, follow-up actions, and outcomes from free-text input.
Editable Interactions: Update only specified fields without overwriting the entire record.
Follow-Up Management: Automatically suggest or adjust follow-up actions.
Material Updates: Track and update materials shared during interactions.
Summary Retrieval: Fetch and summarize recent interactions for quick review.