SalesforceAIResearch/enterprise-deep-research

What it solves

Enterprise Deep Research (EDR) automates complex, multi-step research tasks that require gathering information from diverse sources, analyzing data, and generating reports. It solves the problem of manual, time-consuming research by using an autonomous system that can plan, search, reflect on its findings, and adjust its strategy in real-time.

How it works

EDR uses a multi-agent architecture orchestrated by a "Master Planning Agent." This agent breaks down your research query into smaller tasks and distributes them to specialized agents that search academic, general, and professional databases (like GitHub and LinkedIn). The system includes a reflection mechanism that identifies knowledge gaps and allows for "human-in-the-loop" steering, meaning you can provide feedback to guide the research as it happens. It also features a visualization agent to turn findings into data-driven insights.

Who it’s for

This project is for developers and enterprise teams who need to automate deep-dive data analysis, report generation, and research workflows. It is designed for those who want a flexible, steerable AI assistant that can be integrated into enterprise applications.

Highlights

• Multi-agent orchestration with a Master Planning Agent for adaptive task decomposition.
• Specialized search agents for general web, academic, GitHub, and LinkedIn data.
• Human-in-the-loop steering for real-time research refinement and direction.
• Extensible tool ecosystem supporting document analysis, NL2SQL, and enterprise workflows.
• Multi-provider LLM support including OpenAI, Anthropic, Google, Groq, and SambaNova.
• Built-in visualization agent for creating data-driven insights.
• Includes a full-stack web application with real-time progress streaming.