Skip to main content

Introduction

Welcome! Here you'll find our guides, examples, API reference, and technical blog.

Want to get started ASAP? 🏃🏃‍♀️ Run to our quickstart >

Overview​

AgentStation's API provides virtual workstations that agentic applications can use to complete autonomous tasks.

Think of it like giving AI agents their own laptops.

What is a workstation?​

A workstation is a virtual environment provisioned with different software and hardware depending on your use case (see Workstation Types).

Your agentic application can request a workstation via the API on-demand, then use the browser, meeting and voice action endpoints to take actions.

Once all tasks are completed and any output files are generated successfully, your application can release the active workstation.

AgentStation Overview
The workstation lifecycle

What is a workspace?​

Workspaces contain the file storage and log history associated with Workstations. They can be reused across workstations to facilitate multi-agent workflows.

When you request a workstation, a new workspace is created automatically unless you specify an existing workspace ID.

How to build with AgentStation​

There are two methods to build agentic applications on the platform:

  • Use the AgentStation Agent Framework (Low code): Leverage our in-house state machine to build agents that can drive branching conversations and complete browser tasks. Agents are configured through a JSON file uploaded to the AgentStation platform.
  • Via the API (fully customizable): Utilize our API to build custom workflows to be executed on our virtual workstations. You can leverage our integrated AI models, or hook up your own.

AI Models​

AgentStation has several AI models integrated into the platform:

  • Text-to-speech
  • Speech-to-text (real-time)
  • Speech-to-text (batch)
  • LLM

These models are leveraged in our Agent Framework, as well as some API endpoints. All models are included within our standard pricing, based on agent-minute usage.