Features

Table of Content

Table of Content

Table of Content

Deep Research

Deep Research mode enables Data Agent to go beyond simple lookup or single-pass extraction. It activates a multi-step reasoning agent that transparently shows how it thinks, evaluates sources, and reaches conclusions — giving users visibility into the full chain of thought.

How Deep Research Works

  • The system behaves like an agent, not a black-box LLM.

  • Instead of returning a single answer instantly, it breaks the task into multi-step reasoning chains.

  • Each action the agent takes — from searching, filtering, to validating — is visible in a timeline view.

  • You can see what was searched, why certain results were used or ignored, and how the final output was formed.

When to Use Deep Research

  • You need high-confidence results from complex queries.

  • The answer depends on cross-checking sources or interpreting context.

  • You want transparency into how conclusions are made — not just a single AI guess.

Example Use Cases

  • Researching public risk factors from multiple financial filings.

  • Verifying employee count across multiple pages on a company website.

  • Pulling multi-paragraph summaries from articles or structured databases.

Best Practice Tip

Use Deep Research for columns where accuracy, traceability, or nuance really matter — especially in compliance, finance, and legal analysis workflows.

Columns

Columns

Columns

Formating

Formating

Formating