Skip to content
alfe.ai

Create a managed agent

You create a managed agent from the Alfe dashboard at app.alfe.ai. The flow has two phases: creation (choose a size and confirm billing) and setup (pick a template and configure models). The host is provisioned during setup, so the agent comes online as you finish configuring it.

Before you start

Section titled “Before you start”

You need an Alfe account and a payment method on file. Managed agents draw on your plan — see Choosing a size and the pricing page for current plans and rates.

The flow

Section titled “The flow”
  1. Start a new agent. In the dashboard, create an agent and choose Managed as the hosting type.
  2. Choose a size. Pick small, medium, or large based on how heavy the agent’s workload will be. You can see specs and pricing as you choose — see Choosing a size.
  3. Confirm. Creating the managed agent adds it to your subscription. Once billing is confirmed, Alfe begins provisioning a dedicated host.
  4. Provisioning. Alfe creates the host, issues the agent’s credentials, and brings it online. This typically takes about a minute. You can watch the status change in the dashboard — see Lifecycle & status.
  5. Finish setup. Choose a template (persona, integrations, and starter workspace) and configure which models the agent uses. Every model Alfe offers is available by default; you can set a default and bring your own provider key if you want.

When the agent reports running, it’s connected and ready to receive requests.

What you get

Section titled “What you get”

Every managed agent is provisioned with a set of core integrations already installed — chat, a web console, support, and a headless browser — so it can talk to you and browse the web out of the box. You can add more from the integration registry at any time.

The agent also has its own identity and credentials, its own persistent memory, and its own model configuration. Nothing is shared with your other agents.

After creation

Section titled “After creation”
  • The agent stays online automatically and recovers after a restart.
  • You can start, stop, restart, or recreate it from the dashboard — see Lifecycle & status.
  • You can connect to it and inspect it from a terminal with the CLI.
  • You can drive the whole flow programmatically over MCP or the Agent API.