Skip to content

Data Model in XTM One

XTM One is easier to understand when you treat it as a set of connected objects rather than a set of disconnected screens.

Introduction

At a high level, XTM One works with three kinds of things:

  1. assistants that do work
  2. resources those assistants rely on
  3. records that show what happened

The same object can appear in multiple places in the UI. An agent can appear in Chat, inside a Flow, in History, and in the AI Catalog without becoming four different things.

Core objects

Agents

Agents are the main assistants in XTM One.

An agent can have:

  • instructions and persona
  • tools
  • knowledge bases
  • prompts and skills
  • assignments
  • handovers to other agents
  • memory

If you remember only one object type, remember Agent.

Assignments

Assignments are automation rules attached to an agent.

They answer two questions:

  • what should trigger the work
  • what should the agent do when it runs

Assignments are not separate assistants. They are automation behavior inside an agent.

Runs and history

Runs are the execution records created when assignments or other background work happen.

They let you answer:

  • did the work run
  • is it still running
  • did it fail
  • is it waiting for a person
  • what result was produced

This is the operational layer of the product.

Conversations

Conversations are the chat sessions between a user and the General Assistant or a specific agent.

A conversation keeps the discussion together, while the agent remains the reusable assistant behind it.

Flows

Flows group agents into a larger operating model.

A flow is useful when you need to understand:

  • which agents belong to one workflow
  • which agent is the entry point
  • how handovers work across the group

A flow is not the same thing as an agent. It is a coordination layer above agents.

Reusable resources

The Resources area contains the building blocks that agents and chats depend on.

Knowledge bases

Knowledge bases store documents and indexed content that agents can search.

Prompts

Prompts are reusable text instructions.

Skills

Skills are reusable instruction packages that shape how an agent behaves.

MCP servers

MCP servers expose external tools through the Model Context Protocol.

Tools

Tools can be built-in or custom. They are the actions an agent is allowed to take.

Integrations

Integrations connect XTM One to outside systems.

Access and sharing objects

Some objects exist mainly to control who can see or manage other objects.

Users and groups

Users have personal settings and access. Groups are used for controlled sharing.

Visibility states

Most main objects follow one of these visibility states:

  • personal
  • shared with a group
  • company-managed

Those states matter because they change who can edit the object and whether you should reuse it or make a copy.

How the model appears in the UI

The UI mostly follows this pattern:

  • list page
  • detail page
  • tabs for overview, configuration, activity, and related records

That means the UI is not inventing a new concept on every screen. It is showing different views of the same underlying objects.

Practical example

A simple example looks like this:

  1. A user opens an Agent.
  2. That agent contains one or more Assignments.
  3. The agent references Knowledge Bases, Prompts, Skills, MCP Servers, Tools, and Integrations.
  4. The agent can belong to a Flow.
  5. When the agent or one of its assignments executes, a Run is created in History.
  6. If the user chats with the agent, the interaction is stored in a Conversation.

Once you see that chain, most of the product becomes much easier to read.

Next step

Continue with Relationships and nesting, which explains how these objects point to one another and where configuration actually lives.