> For the complete documentation index, see [llms.txt](https://docs.datumo.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.datumo.com/documentation/documentation-en/get-start/publish-your-docs.md).

# Core concepts and structure

<figure><img src="/files/yWXUR7YDrMhaA91kkmTH" alt=""><figcaption></figcaption></figure>

**Company** is the platform's top-level unit. You can create multiple Workspaces within a single Company, and generally one company or team uses one Company account.

**Workspace** is an independent operating environment for each organization (team, project, business unit, etc.). In each Workspace, you can invite members and set permissions by role (Role), and data and settings do not affect one another between Workspaces.

**Application** is the basic unit of evaluation and monitoring, corresponding to one AI service actually in operation. (e.g., a customer support chatbot, a document summarization service) You can create multiple Applications within a single Workspace.

**Module** is the pipeline component that makes up an Application. By registering each stage, such as RAG, Router, and Generator, as a Module, you can perform two levels of evaluation.

<table><thead><tr><th width="175.8623046875">Evaluation level</th><th width="181.48828125">Target</th><th>Description</th></tr></thead><tbody><tr><td>Module-level eval</td><td>Individual Module</td><td>Measured at the component level, such as RAG retrieval quality or Router classification accuracy.</td></tr><tr><td>E2E eval</td><td>Entire Application</td><td>Verifies the quality of the entire pipeline from user input to final output.</td></tr></tbody></table>

Using both levels in parallel lets you identify the modules responsible for overall quality degradation.