Finsight is an intelligent assistant that automates financial analysis, anomaly detection, and risk profiling using Retrieval-Augmented Generation (RAG), LangChain agents, and local LLMs via Ollama. It parses financial documents, answers company-specific queries, and generates structured PDF reports — empowering analysts, investors, and enterprise decision-makers.
- ✅ Financial Report Generation
- 🔍 Retrieval-Augmented QA (RAG)
- 🚨 Anomaly Detection
⚠️ Risk Report Generation- 📄 PDF Output
- 🤖 LangChain Agent Tools
- 🧪 CLI Interface
| Layer | Technology |
|---|---|
| Language Model | Ollama – mistral |
| Framework | LangChain |
| Retrieval | Chroma Vectorstore |
| Embedding Mgmt | Custom loader + embedding manager |
| PDF Generation | fpdf |
| Interface | Python CLI + LangChain Agents |
| Tool Name | Purpose |
|---|---|
generate_financial_report |
Creates a structured PDF financial report |
retrieval_qa |
Answers user questions using retrieved financial docs |
detect_anomalies |
Detects irregular patterns in company data |
generate_risk_report |
Generates a risk profile and warning report |
finsight/
├── app/
│ ├── cli/ # Command-line interface logic
│ ├── embeddings/ # Vectorstore setup and embedding manager
│ ├── tools/ # Tool definitions for the LangChain agent
│ └── utils/ # PDF and formatting helpers
├── output/
│ └── report/ # Auto-generated PDF reports
├── main.py # CLI Entry point
├── data/ # add regulatory filings for analysis
├── font/
├── requirements.txt
└── README.md
git clone https://github.com/your-username/finsight.git
cd finsight- Create and Activate a Virtual Environment
Copy
Edit
python -m venv venv
source venv/bin/activate # macOS/Linux
venv\Scripts\activate
ollama run mistral
python main.py