AgentComparison
Case (case-computers) · product · persistent computer for AI agents

Case

A hosted persistent computer for developers' own agents, over MCP.

Group: Build-your-own platforms and infrastructure

Case gives an AI agent its own persistent Linux desktop that keeps logins, files and memory between sessions, driven over MCP from tools such as Claude Code, Codex or Cursor. It is not an assistant on its own: you bring the model and the key.

Documentation researched, not testedChecked 1 October 2026Vendor documentation, not a test

The decision in plain English

Developer infrastructure, not a ready-made assistant. Case says one plan costs $29 a month and that you bring your own LLM key. Its code is on GitHub, but the repository licence reads 'Other' and the project is small (3 stars when we checked). Read the licence and what the agent can do on the machine before use.

Best fit

Developers who want an agent to keep logged-in sessions and files between runs, for jobs such as scheduled browser tasks, and who are comfortable running or paying for the machine and supplying their own model key.

Poor fit

Anyone who wants a finished assistant, or who needs a published privacy policy, terms, security audit or support commitments before storing logins on a machine an agent drives.

Documented capabilities

Vendor statements, not verified task outcomes.

  • A persistent desktop whose logins, cookies and files survive sleep (as described) Source · checked 2026-10-03
  • MCP tools such as computer_create, computer_login, computer_sleep and computer_navigate, plus screenshot, exec and files Source · checked 2026-10-03
  • Human handoff for 2FA, captchas and approvals by phone link Source · checked 2026-10-03
  • Saved skills and scheduled recurring runs on the machine Source · checked 2026-10-03

Permissions and human control

  • Case says credentials are saved once by a human through a one-time link, encrypted, and typed into the site's login page without the agent or API seeing the password. It says 2FA codes, captchas and approvals pause the run and ask a human by phone link. These are vendor statements; the agent can still act inside the logged-in accounts on that machine.

Documented controls do not prove that an agent always follows them.

Limits and open questions

  • No privacy policy or terms page found on the site
  • Licence shown as 'Other' on GitHub; read it before commercial use
  • Small early project: 3 stars, one listed contributor at the time we checked
  • You pay your own LLM costs on top of the $29 plan
  • Described by the vendor; not tested by AgentComparison

Privacy and security

Case says logins are kept in an encrypted vault and not shown to the model, and the machine keeps cookies and files between sessions. We found no privacy policy, terms page or independent audit. Anything saved on the machine persists, so treat it like handing an agent a logged-in computer.

Common questions about Case

What is Case for?

Developers who want an agent to keep logged-in sessions and files between runs, for jobs such as scheduled browser tasks, and who are comfortable running or paying for the machine and supplying their own model key. The vendor describes it as: Case gives an AI agent its own persistent Linux desktop that keeps logins, files and memory between sessions, driven over MCP from tools such as Claude Code, Codex or Cursor. It is not an assistant on its own: you bring the model and the key.

Who should look elsewhere?

Anyone who wants a finished assistant, or who needs a published privacy policy, terms, security audit or support commitments before storing logins on a machine an agent drives.

How much does Case cost?

case.computer lists one plan at $29 a month: a persistent computer, MCP access, console, scheduled runs and 2FA handoffs, cancel anytime with a 7-day refund on the first month. Self-hosting from GitHub is also offered; we did not verify what it costs to run.

What can it access and what needs my approval?

Case says credentials are saved once by a human through a one-time link, encrypted, and typed into the site's login page without the agent or API seeing the password. It says 2FA codes, captchas and approvals pause the run and ask a human by phone link. These are vendor statements; the agent can still act inside the logged-in accounts on that machine.

What happens to my data?

Case says logins are kept in an encrypted vault and not shown to the model, and the machine keeps cookies and files between sessions. We found no privacy policy, terms page or independent audit. Anything saved on the machine persists, so treat it like handing an agent a logged-in computer.

Has AgentComparison tested it?

No. This record is built from the vendor's own documentation, checked 2026-10-03. Vendor claims are attributed, not confirmed.

How to evaluate Case yourself

  1. Pick one low-risk task that matches its documented capabilities: a persistent desktop whose logins, cookies and files survive sleep (as described).
  2. Write the expected output and what counts as failure before you start.
  3. Connect the minimum access needed, and check approval settings first.
  4. Run it twice and compare, then review every action it took.
  5. Check cost after any free allowance. Unknown price is not free.

Related

Documented use cases: Building your own agent.

Sources and history

Change history

Documentation overview expanded with current pricing/access sources. No hands-on results claimed.

Report a correction