# LangAlpha

An agent harness for interpreting financial markets that gives an agent a persistent research workspace per goal, so analysis and theses compound across sessions instead of resetting each time.

LangAlpha is an agent harness built to interpret financial markets and support investment research, built on LangChain and LangGraph. Rather than treating investing as a single question-and-answer exchange, it gives each research goal its own persistent workspace: the agent interviews the user about their goals and style, produces an initial deliverable, and keeps building on saved files, threads and research the next time the user returns.

It can pin a curated news brief, dispatch parallel subagents to screen the market, and present ideas in an interactive dashboard. Tools are loaded progressively, as summaries first and full documentation only when a tool is actually used, and the agent can write and run Python against a sandbox to process financial data from MCP servers instead of pouring raw data into the model's context window. It ships web, terminal and API interfaces.

Website: https://langalpha.ai