Breaking

Renmin University Open-Sources FinSight, Generating Financial Reports From a Ticker

July 1, 2026 at 04:30 EDT

AI Agents
Open Source
Finance

Renmin University of China (RUC-NLPIR) has released FinSight, an open-source multi-agent system that automatically generates comprehensive financial research reports of over 20,000 words from a single ticker symbol. The project is available on GitHub and has drawn attention among developers who curate AI agent tooling.

RUC-NLPIR · Open Source · ACL 2026 Main

FinSight: One Stock Ticker In, a 20,000-Word Research Report Out

An open-source multi-agent system that automates the full analyst pipeline — collecting data, running analysis, generating charts and producing publication-ready reports from a single ticker like AAPL or 000001.

20,000+

words per generated report

markets: A-shares, HK, US

output formats: DOCX / PDF

From one ticker to a finished report

The system runs the full analyst workflow end to end.

Ticker In

AAPL · 000001

→

Collect Data

statements · prices · macro

→

Analyze + Chart

CoA · VLM check

→

Report Out

DOCX / PDF

Three core mechanisms

CAVM

Code Agent with Variable Memory — unifies data, tools and agents in one variable space via code execution.

Vision-Enhanced

A VLM iteratively fixes chart flaws — missing legends, scale errors — before publishing.

Two-Stage Writing

Builds a structured Chain-of-Analysis (CoA) first, then drafts the full report.

What the paper reports

Substantially outperforms existing deep-research systems on company- and industry-level tasks in factual accuracy, analytical depth and presentation quality. Accepted to ACL 2026 Main.

Still to be proven

Released early, so real-world assessments remain limited. Roadmap: expanded US coverage, a plugin system and stronger sandboxing — to be tested through forks and issues.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.