Skip to content
alfe.ai

Managed agents

A managed agent runs on infrastructure that Alfe provisions and operates for you. You don’t supply a machine, install anything, or keep a process alive — Alfe stands up a dedicated host for the agent, connects it to the platform, and keeps it running around the clock.

This is the fastest way to get a durable, always-on agent. You create it from the dashboard, pick a size, and it’s online in about a minute.

Managed vs self-hosted

Section titled “Managed vs self-hosted”

Every agent is one of two hosting types:

Managed Self-hosted
Where it runs On an Alfe-provisioned host On your own machine or server
Who keeps it online Alfe (24/7, auto-restart) You
Setup Create it in the dashboard Run alfe setup yourself
Compute cost Included in your plan / add-ons Your own hardware
Lifecycle controls Start, stop, restart, recreate You manage the process

Both hosting types are full agents: each has its own identity, memory, credentials, integrations, and model configuration. The only difference is where the runtime lives and who keeps it alive. See Core concepts for the shared building blocks.

When to use a managed agent

Section titled “When to use a managed agent”

Choose managed when you want:

  • An agent that’s online 24/7 without running your own server.
  • Automatic recovery — the host restarts the agent on reboot or crash.
  • No local setup, dependency installs, or process supervision.
  • A clean, dedicated environment per agent.

When to use a self-hosted agent

Section titled “When to use a self-hosted agent”

Choose self-hosted when you want:

  • The agent to run on hardware you already control.
  • Direct local access to files, networks, or devices on your own machine.
  • To develop and test an agent on your laptop before making it always-on.

You connect a self-hosted agent with the CLI — see alfe setup.

What’s next

Section titled “What’s next”