CallSense

CallSense is an agentic AI system for call intelligence, QA, and operational routing.

Features

• Audio or transcript intake
• OpenAI Whisper transcription
• Structured summaries and QA scoring with Pydantic
• Deterministic routing decisions
• Streamlit UI with audit trail

Quickstart

1. Create and activate a virtual environment:

   python -m venv .venv
   .venv\\Scripts\\activate

2. Install dependencies:

   pip install -r requirements.txt

3. Create a .env file (copy from .env.example) and add your OpenAI key:

   copy .env.example .env

   Then edit .env and set:

   • OPENAI_API_KEY (required)
   • CALLSENSE_OPENAI_MODEL (optional, default: gpt-4o-mini)
   • CALLSENSE_WHISPER_MODEL (optional, default: whisper-1)
   • CALLSENSE_REQUEST_TIMEOUT (optional, default: 60)

4. Run the app:

   streamlit run app.py

   Browse and upload call center calls,in home screen, available in /resources folder and get the call sense analysis

Architecture

Streamlit UI
     ↓
LangGraph Orchestration
     ↓
Call Intake Agent
     ↓
[ Audio? ] → Transcription Agent
     ↓
Summarization Agent
     ↓
QA Scoring Agent
     ↓
Routing Agent
     ↓
Results + Actions + Audit

Project Layout

• app.py Streamlit UI
• callsense/ core package
  • agents/ single-responsibility agents
  • graph/ LangGraph workflow
  • models/ Pydantic schemas
  • llm/ LLM factory
  • utils/ audit helpers