---
title: "Machines & Fleet"
canonical: https://trickest.com/docs/key-concepts/machines-and-fleet
description: "Understand fleets and worker machines for Trickest workflow execution, and how they differ from managed agent sandboxes."
---

# Machines & Fleet

## Overview

Choose a fleet when you run a workflow. Its machines execute the jobs, and its
capacity determines how many jobs can run at once.

## What are Machines and Fleets

**Machines** are the compute resources that execute your workflows. Every run needs at least one machine; each machine executes one workflow node's job at a time.

A **fleet** is a pool of machines. The execution engine assigns jobs to available
machines in the fleet you select.

This page describes **workflow execution machines**. An [agent sandbox](/docs/using-the-app/ai-agent/sandboxes)
is the agent's working environment for commands, files, and apps. Its lifecycle
is separate from a workflow node's job. The fleet availability table below does
not specify standalone sandbox access or pricing.

## Fleet Types

| | Managed | Self-Hosted |
|---|---|---|
| **Machines provided by** | Trickest | You |
| **Setup required** | None | Install the Trickest Agent on your machines |
| **Scaling** | Trickest manages capacity | Limited to the machines you attach |
| **Availability** | Enterprise plans | All plans |

**Managed fleets** are pools of virtual machines that Trickest provisions, scales, and maintains for you. You run workflows on them without any infrastructure setup. Managed fleets are available on Enterprise plans. See [Managed Machines](/docs/key-concepts/managed-machines).

**Self-hosted fleets** use your servers, VMs, cloud instances, or macOS devices.
Install the Trickest Agent on each device to attach it to the platform. Workflows
then run on your infrastructure, with access to your network and hardware.
Self-hosted fleets are available on all plans. On the
[Community plan](/docs/key-concepts/community-plan), Trickest creates a self-hosted
fleet at signup for all workflow runs. See
[Self-Hosted Machines](/docs/key-concepts/self-hosted-machines).

## How a Run Uses a Fleet

The execution engine creates node jobs and assigns each to an available machine
in the selected fleet. Each machine runs one job at a time. Parallel branches and
[distributed nodes](/docs/using-the-app/workflow-and-executions/distributing-and-scaling-jobs)
spread work across multiple machines, up to the fleet's machine cap.

If more jobs are pending than the fleet has machines, jobs queue and start as machines free up. Multiple runs can target the same fleet at the same time and share its pool.

## How It Relates

* **[Workflows](/docs/key-concepts/workflows):** Define the nodes and their connections.
* **Runs:** Execute the workflow on a fleet. Inspect job status to see queued, running, or finished work.
* **[Solutions](/docs/key-concepts/solutions-database):** Run their workflows on the fleet you choose.

## Next Steps

<CardGroup cols={2}>
  <Card title="Managed Machines" icon="cloud" href="/docs/key-concepts/managed-machines">
    Trickest-provided compute with automatic scaling.
  </Card>
  <Card title="Self-Hosted Machines" icon="server" href="/docs/key-concepts/self-hosted-machines">
    What the Trickest Agent is and what it can access on your machines.
  </Card>
  <Card title="Distributing and Scaling Jobs" icon="diagram-project" href="/docs/using-the-app/workflow-and-executions/distributing-and-scaling-jobs">
    Split file and folder outputs into multiple parallel jobs.
  </Card>
</CardGroup>

---
_Markdown source of https://trickest.com/docs/key-concepts/machines-and-fleet._