Contextly

Contextly.ai turns talk into action — bridging human conversations with intelligent automation.

Contextly.ai transforms client–advisor conversations into actionable tasks using AI. It provides a React + Vite frontend and a FastAPI backend that connects to Hugging Face’s Apertus 8B Instruct model.

🚀 Features

Frontend (React + Vite)

1. Upload or select transcript samples
2. View raw and cleaned transcripts with highlighted evidence
3. Extracted tasks and todos from client conversations

Backend (FastAPI)

1. POST /extract_labels → classify transcript
2. GET /list_transcripts → list all available .txt
3. GET /get_transcript/{filename} → get transcript text

Swagger UI available at http://127.0.0.1:8000/docs

AI Model

Apertus 8B Instruct 2509 via Hugging Face Router API Optimized for high recall in compliance-sensitive banking tasks

Coming soon...

1. Extracts 8 canonical task types with evidence spans, highlight text where tasks are extracted from within transcript
2. Send Feeback to the Model
3. Improve chatbox

⚙️ Backend Setup (FastAPI)

1. Clone & enter backend cd backend

2. Create and activate virtual environment For macOS/Linux: python3 -m venv .venv source .venv/bin/activate For Windows: python -m venv .venv .venv\Scripts\activate

3. Install dependencies python -m pip install -r requirements.txt

4. Run FastAPI server Development: uvicorn app.main:app --reload --port 8000

Deployment: uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload

5. Test backend & API Documentation Open http://127.0.0.1:8000/docs to try endpoints directly.

🎨 Frontend Setup (React + Vite)

1. Clone & enter frontend cd frontend

2. Install dependencies npm install

3. Run dev server npm run dev

Runs at http://localhost:5173.

Frontend will call /extract_labels → backend → Hugging Face model.

🌐 API Endpoints

POST /extract_labels → classify transcript

GET /list_transcripts → list all available .txt

GET /get_transcript/{filename} → get transcript text

🔒 Environment Variables

Backend requires Hugging Face API token: HF_TOKEN=your_hf_token_here

Put this in backend/.env.

📊 Tech Stack

Frontend: React, Vite, TypeScript, CSS Backend: Python, FastAPI, Uvicorn AI Model: Hugging Face Apertus 8B Instruct Deployment: Azure