Skip to main content

​
Sign in

Go to labframe.ai and click Get started. Enter your email address — LabFrame will send you a magic link. Click the link in your inbox and you are signed in. No password needed.
If your organization uses SSO, you may be redirected to your university or company login page instead.

​
Create an agent

Click + New Agent in the sidebar. Enter a name — something like “Research Assistant” or “Course Helper” — and select a machine type. Click Create. LabFrame provisions a dedicated cloud VM for your agent. You will see status messages while the machine starts up:
  • Creating VM…
  • Starting VM…
Once the agent appears in the sidebar, it is ready. It now has its own computer with a full Linux environment, persistent filesystem, and installed tools.

​
Send your first message

Type a request in the chat input at the bottom of the screen. For example:
Survey the top 5 recent papers on transformer efficiency and write a one-page summary as a PDF.
Press Enter to submit. Your agent starts working on its own machine:
  • A thinking indicator while the agent plans its approach
  • Tool activity summaries as it searches the web, writes files, runs commands (collapsed by default — click to expand)
  • The final response with text and file links
The agent does everything on its dedicated computer — the same way a human researcher would work on their laptop, except it does it in minutes.

​
Open a deliverable

When the agent produces a file, it appears as a clickable link in the chat message.
  • Click a .webapp link to open a deployed website in a new tab
  • Click a .pdf link to preview the PDF inline
  • Click any other file to open it in the workspace file preview
Switch to the Workspace tab to browse everything the agent has created on its computer — source files, build output, data, and deliverables.

​
Automate a larger task

For work that takes multiple steps, use Goal Mode. Click Set goal below the chat input, describe the objective and acceptance criteria, and let the agent run autonomously. It plans, executes, and iterates on its own computer until the goal is met.
Goal: Produce a comparative analysis of 3 battery degradation models. Acceptance criteria: PDF report with methodology comparison table, at least 10 cited sources, and a recommendations section.
The agent works in the background. Check back when it’s done — all the files will be in the workspace.

​
What’s next

Agents

Dedicated VMs, system prompts, models, and agent states.

Goal Mode

Delegate complex tasks and let the agent work autonomously.

Web Apps

How agents build and deploy websites on their own machines.

Collaboration

Invite others to work with your agent.