DeepAnalyze
An agentic large language model for autonomous data science that plans, analyzes, models, visualizes and writes analyst-grade reports over structured, semi-structured and unstructured data sources without human intervention.
DeepAnalyze is described as the first agentic large language model for autonomous data science, built to complete data-centric tasks including data preparation, analysis, modeling, visualization and report generation without human intervention. It can research diverse data sources, including structured databases, CSV and Excel files, semi-structured JSON, XML and YAML, and unstructured text and Markdown, and produce analyst-grade research reports.
The model, code, training data and a demo are all open-sourced, so a team can deploy or extend its own data-analysis assistant. Companion projects from the same authors include SkillAdam, a plugin that automatically improves the skills of coding agents from a SKILL.md file, EvoOntology, a self-evolving ontology layer for data agents, and DeepPrep, an agentic system for autonomous data preparation.