> For the complete documentation index, see [llms.txt](https://docs.autommate.com/userguide/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.autommate.com/userguide/start-here/architecture-at-a-glance.md).

# Architecture at a glance

Autom Mate connects the way a request arrives with the work needed to fulfil it. Use this map to understand where Hyperflows, AI Agents, Service Automation and execution runtimes fit together.

## The platform and the runtime

**Autom Center** is the central platform where teams build, configure and monitor their automations. **Mate Agents** execute automation work and connect to the systems and dependencies that work requires. You use a browser to manage the platform; the selected runtime determines where the work executes.

```mermaid
flowchart TB
    U["People and administrators"] --> C["Autom Center<br/>Build · configure · operate"]
    C --> H["Hyperflows<br/>Actions and data mappings"]
    C --> S["Service Automation<br/>Blueprints and deployments"]
    S --> W["Released Work Items<br/>Reusable Hyperflows"]
    H --> R["Mate Agent<br/>Execution runtime"]
    W --> R
    R --> T["Connected applications<br/>APIs · files · scripts"]
    R --> L["Execution records and logs"]
    S --> E["Service Cases and Pulse"]
    L --> C
    E --> C
    classDef platform fill:#e8f3ff,stroke:#2463a6,color:#152b43;
    classDef execution fill:#e2f6f1,stroke:#187564,color:#123e35;
    class C,H,S,W platform;
    class R,T execution;
```

This is a logical product map. Deployment location and network connections are described in [Cloud, hybrid and on-premises deployment](/userguide/administration-and-runtimes/deployment-options.md).

## Two ways AI uses automation

| Entry point                        | What you configure                                  | How it reaches execution                                                                 |
| ---------------------------------- | --------------------------------------------------- | ---------------------------------------------------------------------------------------- |
| **AI Agent Composer**              | Model, instructions, knowledge, skills and channels | The agent uses its configured skills to perform the supported work                       |
| **External AI client through MCP** | Client connection and Autom Mate authorization      | The client discovers exposed tools and can run supported existing API/webhook Hyperflows |

```mermaid
flowchart LR
    A["AI Agent Composer<br/>Knowledge · skills · channels"] --> K["Configured skills"]
    K --> H["Automation work"]
    M["External AI client"] --> P["Autom Mate MCP<br/>Authorized exposed tools"]
    P --> X["Discover · inspect<br/>Run supported existing Hyperflows"]
    X --> H
```

An **AI Agent** is the conversational assistant. A **Mate Agent** is the execution runtime. They have different configuration and responsibilities. In Dublin 6.0.0, the MCP tool set does not create or deploy Hyperflows or manage Service Automation. Start with the [MCP connection guide](/userguide/mcp-and-ai-reference/connect-an-ai-client-with-mcp.md) for the supported path.

## Add a service around reusable work

A Hyperflow defines automation logic. A Work Item packages reusable Hyperflow work for a service phase. A service blueprint defines an outcome and its **Engage → Govern → Execute → Prove** lifecycle. A deployment selects the version, runtime and configuration used for a particular target. Service Cases record individual requests.

For example, a provider can define one access-request service and configure deployments for different customer environments. The reusable definition stays separate from the selected connections and inputs. See [MSP Edition](/userguide/service-automation/msp-edition-overview-and-first-service.md).

Use [the service model](/userguide/service-automation/understand-the-service-model.md) to design the outcome. Use [runtime setup](/userguide/administration-and-runtimes/plan-execution-and-runtime-setup.md) to choose the execution host. A platform installation and a service deployment are separate tasks.

## Choose the next step

* Place the components: [deployment options](/userguide/administration-and-runtimes/deployment-options.md).
* Prepare customer-managed infrastructure: [system requirements](/userguide/administration-and-runtimes/system-requirements.md) and [networking](/userguide/administration-and-runtimes/network-and-connectivity.md).
* Build one working automation: [Build a Hyperflow](/userguide/build-automations/build-a-hyperflow.md).
* Review data access and responsibilities: [security and data handling](/userguide/administration-and-runtimes/security-and-data-handling.md).