This project is an experimental prototype and is still being actively developed.
Features, prompts, and tool integrations may change frequently.
An AI-powered multi-agent system built with LangGraph and Gradio for analyzing competitor product launches.
The assistant can perform three types of analysis:
- π Competitor Analysis β positioning, strengths, weaknesses, strategic takeaways
- π¬ Market Sentiment β positive/negative signals from social media, reviews, forums
- π Launch Metrics β KPIs, adoption, press coverage, traction insights
It uses OpenAI GPT models with a tool-augmented workflow (Playwright browser, web search, Firecrawl, etc.) to provide fresh, evidence-backed reports.
product_intel/ βββ agents.py # Analyst role prompts (competitor, sentiment, metrics)
βββ tools.py # External tools: Playwright, Serper, Wikipedia, Firecrawl, Python REPL
βββ graph.py # LangGraph Sidekick orchestration (worker, tools, evaluator loop)
βββ ui.py # Gradio UI (chatbot + buttons for analysis modes)
βββ app1.py # Entry point to launch the app
- Multi-role prompts (Competitor / Sentiment / Metrics)
- Tool-augmented reasoning with:
- Playwright (async browsing & scraping)
- Google Serper (web search)
- Wikipedia
- Python REPL
- File sandbox
- Push notifications (Pushover)
- Firecrawl (optional deep crawl/search)
- Evaluator loop β ensures output meets success criteria, or asks clarifying questions
- Gradio interface with persistent chat history and one-click mode selection
git clone https://github.com/yourusername/product-intel-sidekick.git
cd product-intel-sidekick
uv sync # or pip install -r requirements.txt
2. Install Playwright browsers
playwright install
3. Configure environment variables
Create a .env file in the project root:
OPENAI_API_KEY=your_openai_api_key_here
SERPER_API_KEY=your_serper_api_key # optional, for Google search
FIRECRAWL_API_KEY=your_firecrawl_key # optional, for Firecrawl
PUSHOVER_TOKEN=your_pushover_token # optional, for notifications
PUSHOVER_USER=your_pushover_user # optional, for notifications
βΆοΈ Run the app
uv run product_intel/app1.py
π₯οΈ Usage
Enter a company name or product in the chat.
Choose an analysis type:
π Competitor Analysis
π¬ Market Sentiment
π Launch Metrics
The assistant will gather evidence (via search/crawl) and generate a structured report.
You can refine the query or ask follow-up questions directly in chat.
π¦ Requirements
Core libraries:
langgraph
langchain
langchain-openai
gradio
playwright
python-dotenv
requests
π οΈ Development Notes
Modular design: prompts (agents.py), tools (tools.py), orchestration (graph.py), UI (ui.py)
Easy to extend: add new tools in tools.py or new analyst roles in agents.py
Compatible with both Gradio (default) and Streamlit (if you want to port later)