Finance is an open-source AI assistant designed to bring institutional-grade financial intelligence to developers and analysts through natural language interaction. Built on top of Valyu’s unified search API, it aggregates live market data, SEC filings, academic research, and real-time news into a conversational interface. Unlike traditional financial platforms that require expensive subscriptions or fragmented tools, Finance consolidates these capabilities into a single chat-based workflow. It’s ideal for quantitative analysts, hedge fund researchers, and fintech developers who need deep financial insights without switching between multiple proprietary platforms. The application supports both cloud-based and fully private, self-hosted modes, making it suitable for sensitive research or offline environments.
The system is powered by a combination of Valyu’s specialized financial data API, Daytona for secure Python code execution in sandboxes, and local LLMs via Ollama or LM Studio — enabling unlimited, private queries without API costs. With no authentication required in self-hosted mode and SQLite-based persistence, Finance offers a developer-friendly experience for prototyping financial AI applications without vendor lock-in or billing constraints.
What You Get
- Institutional-Grade Financial Data - Access live global market data, SEC filings (10-Ks, 10-Qs, 8-Ks), insider trading reports, and academic research through Valyu’s unified API.
- Secure Python Code Execution - Run complex financial models, backtests, and ML algorithms in isolated Daytona sandboxes with real-time output and error handling.
- Interactive Visualizations - Generate publication-ready charts and dashboards using Recharts with automatic data aggregation from multiple sources.
- Local LLM Support - Run unlimited, private queries using Ollama or LM Studio without cloud APIs — with automatic model detection and switching in the UI.
- Self-Hosted Mode - Fully offline operation with SQLite database storage, mock authentication, and no rate limits — ideal for development and sensitive analysis.
- Real-Time News & Sentiment Analysis - Integrate live web search to assess market impact of breaking news and social sentiment on financial instruments.
- Multi-Source Research Aggregation - Automatically combine data from SEC filings, academic papers, news, and market data to answer complex financial questions.
Common Use Cases
- Building a private quantitative research platform - Analysts use Finance to run Monte Carlo simulations on stock prices using Python in Daytona sandboxes, with results visualized and saved locally without exposing proprietary strategies to cloud providers.
- Analyzing SEC filings for hedge fund due diligence - Users query 10-Ks to extract key financial ratios, management commentary, and risk factors using natural language — eliminating manual PDF review.
- Developing AI-powered financial tools without API costs - Developers self-host Finance with Ollama to avoid OpenAI or Valyu billing, enabling unlimited queries for internal tools and prototyping.
- Teaching financial data analysis in academic settings - Professors deploy Finance locally to give students access to real-world financial data and code execution without licensing fees or cloud dependencies.
- Monitoring market sentiment after corporate events - Teams track how CEO statements or earnings calls impact stock prices by combining news, insider trading data, and historical volatility in a single query.
- DevOps teams building private AI agents for finance - Engineers integrate Finance’s architecture (Valyu + Daytona + Ollama) as a reusable stack for custom financial AI agents requiring data, computation, and privacy.
Under The Hood
This project is a Next.js-powered financial analytics platform that merges enterprise-grade authentication with local AI inference capabilities, offering flexibility for both cloud and self-hosted deployments. It supports hybrid database configurations and integrates seamlessly with Valyu’s OAuth ecosystem and Supabase for backend operations.
Architecture
This application adopts a modern monolithic structure built around the Next.js App Router, enabling scalable frontend development with clear separation between UI and backend logic.
- Modular organization with distinct directories for app, components, lib, and utils
- Clear separation between frontend UI components and backend API handlers
- Use of Supabase for authentication and database operations with local SQLite fallback
Tech Stack
Built on TypeScript and Next.js, the project leverages a modern React ecosystem with extensive third-party integrations.
- TypeScript for type safety and enhanced developer experience
- Next.js 14 with App Router for modern React rendering and server-side rendering
- Supabase for authentication and database integration, alongside Ollama/LM Studio support for local AI inference
- Tailwind CSS with Drizzle ORM and Puppeteer for UI and document generation tools
Code Quality
The codebase reflects a balanced approach to structure and maintainability, with consistent use of TypeScript and error handling practices.
- Comprehensive error handling throughout API endpoints with detailed logging
- Type-safe implementation using TypeScript interfaces and types
- Modular component design with reusable UI elements and utility functions
- Security-focused practices including input validation and sanitized error responses
What Makes It Unique
This platform distinguishes itself through its hybrid architecture and flexible deployment options tailored for enterprise use cases.
- Unique dual-model approach supporting both cloud and local AI inference with automatic fallback
- Comprehensive OAuth integration with Valyu platform and custom session management
- Hybrid database support allowing local SQLite and Supabase database flexibility
- Self-hosted mode with configurable AI providers for enterprise deployment scenarios