Investment research is being automated. Dexter, a new open-source autonomous agent released on GitHub, is designed to do the deep financial analysis work that once required a team of analysts — breaking down complex financial questions into structured research plans, executing them against live market data, and self-validating the results before presenting them.
What Is Dexter?
Dexter describes itself as “Claude Code for financial research.” Built in TypeScript and powered by the Bun runtime, it is an autonomous agent that takes broad financial questions — “Is NVIDIA fairly valued given its current cash position?” or “What are the key risks in Apple’s most recent 10-K?” — and turns them into step-by-step research tasks that it executes independently.
Key Capabilities
Intelligent Task Planning
Dexter doesn’t just retrieve data — it decomposes complex financial queries into a structured research workflow. Before touching any API, it plans which data sources to consult, in what order, and how to synthesize the findings. This planning layer is what separates it from simple financial data dashboards.
Autonomous Execution with Real-Time Market Data
Dexter connects to live financial data sources including income statements, balance sheets, and cash flow statements. It uses Financial Datasets API (financialdatasets.ai) for institutional-grade market data — with free access to AAPL, NVDA, and MSFT — and Exa API for web search augmented with real-time news and filings.
Self-Validation and Iteration
Perhaps the most impressive feature is Dexter’s self-check loop. After producing initial findings, the agent evaluates its own work, identifies gaps or contradictions, and re-runs tasks until it reaches a confident, data-backed conclusion. Built-in loop detection and step limits prevent runaway execution — a critical safety feature when autonomous agents are making decisions about where to look for data.
Claude Code-Inspired Design
The interface will feel immediately familiar to developers who’ve used Anthropic’s Claude Code. Dexter’s interactive mode presents a running log of its thinking: which task it’s tackling, what tools it’s calling, and what intermediate conclusions it’s drawing. This transparency is rare in financial AI products and makes the system trustworthy for serious analysis work.
Architecture: TypeScript + Bun
Choosing TypeScript for a financial research agent is a deliberate decision. Unlike Python-dominated AI research, TypeScript brings type safety to a domain where data structures (financial statements, ticker symbols, time series) are highly structured. Combined with the Bun runtime — significantly faster than Node.js for I/O-bound workloads — Dexter achieves snappy, responsive analysis sessions.
The evaluation suite uses LangSmith for tracking and an LLM-as-judge approach for scoring correctness, making it easy to benchmark improvements over time.
Getting Started
Dexter requires a Bun runtime (v1.0+), an OpenAI API key, and a Financial Datasets API key. Optional but recommended: an Exa API key for enhanced web search. Installation is straightforward:
git clone https://github.com/virattt/dexter.git
cd dexter
bun install
cp env.example .env
# Edit .env with your API keys
bun startWhatsApp Integration
Perhaps the most consumer-friendly feature: Dexter can be connected to WhatsApp, allowing users to text a financial question and receive a structured research response directly in their chat app. This brings institutional-grade financial research to anyone with a smartphone and an API key.
Why It Matters
Retail investors and independent analysts have historically been at a significant information disadvantage relative to institutional players with analyst teams. Dexter — and agents like it — represent a meaningful democratization of financial research capability. While it’s not a replacement for human judgment, it dramatically lowers the cost of answering complex financial questions thoroughly. In a market where information advantage is everything, open-source agents like Dexter are a genuine game changer.
Featured image: Financial data charts and market analysis representing AI-powered investment research.