What is DataFact?

DataFact is an AI analyst for documents and databases, built for answers you can verify. It gives teams one workspace for connecting structured data and source documents, organizing reusable business knowledge, and using configured Agents to answer questions or produce supported outputs.

DataFact separates source access, usable analytical resources, reusable knowledge, and Agent behavior. A database connection gives the workspace a path to discover schemas. The Document Library organizes uploaded files. MemoryLake resources make selected database or document material available to compatible workflows. Knowledge and versioned Analysis Models add reviewed terminology, relationships, examples, and business guidance without changing the source records themselves.

An answer may include citations, query context, charts, generated files, or execution details when the Agent and workflow produce them. These artifacts make parts of the result easier to inspect. They do not guarantee that the AI chose the right source, applied the right definition, or reached a correct conclusion. Source availability also depends on configuration, permissions, indexing, and the work performed for the question.

Why it matters

Business questions frequently cross measurable database facts and context stored in documents. Keeping those resources, the applied business context, and the analytical workflow distinct makes it easier to configure access and review an important result.

How DataFact uses its main objects

What you can do

• Upload and organize source files in the Document Library.
• Connect supported databases.
• Create SQL Datasets, Documents (Lite), and Documents (Excel) in MemoryLake.
• Maintain Dataset knowledge and versioned Analysis Models.
• Ask Agents questions and inspect returned evidence, charts, or files when an answer provides them.
• Use Preview reporting and evaluation workflows when they are available.

What DataFact does not guarantee

An AI-generated answer can be incomplete or incorrect. A connected source does not guarantee that every answer uses it, and a citation does not by itself prove that the interpretation is correct. Verify important conclusions against source data.

Example

An Agent investigating a supplier variance might query a Dataset for delivery performance and retrieve a contract clause from a document resource. The reviewer can inspect the supporting artifacts that the workflow returns, but must still confirm that the supplier, date range, clause, and metric definition match the question.

Related concepts

• Documents and databases
• Evidence and citations
• Datasets and Analysis Models
• Agents, Sessions, and messages
• Knowledge and answer quality

Choose a starting point

• For database analysis, connect a database and analyze data.
• For tabular files, import and analyze an Excel document.
• For document retrieval, build a document knowledge experience.