ClickPost Intent Dashboard

An end-to-end AI pipeline that identifies D2C brands showing buying intent for ClickPost's post-purchase platform, scores them by intent strength, and generates personalized outbound sequences. Includes a clean, modern Streamlit UI for reviewing scores, signals, and outreach messaging.

Features

• Signal Extraction: Pulls open roles (via Greenhouse/Lever ATS), public complaints (Tavily Search/Reddit), tech-stack competitors (subdomain resolution & HTML scanning), and growth/leadership news (Google News RSS).
• Intent Scoring: Uses a heuristic algorithm to assign a confidence score (0-100) based on signal weights and data freshness.
• AI Outbound Generation: Leverages Groq (Llama 3.3) to dynamically write personalized LinkedIn messages and follow-up/breakup email sequences tailored strictly to the collected signals.
• SaaS Dashboard: A premium, high-density Streamlit application mimicking a production CRM environment.

Setup

1. Install dependencies

pip install -r requirements.txt

2. Configure API Keys

Create a .env file in the project root:

GROQ_API_KEY=your_groq_api_key_here
TAVILY_API_KEY=your_tavily_api_key_here

3. Run the Backend Pipeline

Generate the data by running the pipeline script. Outputs are saved as JSON files in the output/ directory.

# Run for all accounts
python main.py

# Quick test with only 3 accounts
python main.py --accounts 3

# Test with a specific brand
python main.py --brand "Vuori"

4. Run the Dashboard UI

Start the Streamlit interactive dashboard:

streamlit run app.py

Project Structure

├── app.py                          # Streamlit UI Dashboard
├── main.py                         # Backend data collection & scoring pipeline
├── config.py                       # Configuration (brand lists, keywords, weights)
├── collectors/                     # Signal extraction modules
│   ├── greenhouse_collector.py     # ATS API collector
│   ├── techstack_collector.py      # Subdomain & HTML analysis
│   ├── news_collector.py           # Google News RSS fetcher
│   └── review_collector.py         # Tavily / Reddit complaint discovery
├── scoring/
│   └── intent_scorer.py            # Signal aggregation and heuristic scoring
├── activation/
│   └── outbound_generator.py       # Groq-powered messaging prompt chains
└── output/                         # Generated JSON artifacts (scores, raw signals, outbound)