House dark · evalab.lifestyle
Real world sandboxes for testing and training AI agents.
Deploy simulation environments with stateful twins of the APIs, CLIs, and MCPs your agents use.
Twin Runs
Spin up short-lived or permanent service twins without deploying app code.
Expires automatically after 60 minutes.
Ready · Twin run RUN-2048 is lit. 58 minutes remaining.
Trusted by engineering teams — we do not invent a logo row. The house is unassigned until you book a seat.
Four cues
Run the real world on demand.
Connect twins by changing one line of code.
Start from your system
Deploy your agent in staging, a preview server, or locally. The black box holds the understudies — you keep the lead.
Twins · Twin Runs · Scenarios · Tests · Integrations · MCP · Settings
No app deploy required to stand up a twin. Pick services, set a session, strike the ghost light when you are done.
Environment ready
Twin run · RUN-2048 · Support agent sandbox · 58 minutes remaining · Ready
| Service | Endpoint | Status |
|---|---|---|
| Slack | https://slack.run-2048.twins.evalab.lifestyle | Serving |
| GitHub | https://github.run-2048.twins.evalab.lifestyle | Serving |
| Google Calendar | https://gcal.run-2048.twins.evalab.lifestyle | Serving |
| Stripe | https://stripe.run-2048.twins.evalab.lifestyle | Serving |
Connect the twins
Change one base URL to connect Evalab twins, then seed scenarios with mock or production data from your services.
$ open .env SLACK_API_BASE_URL = https://slack.com/api GMAIL_API_BASE_URL = https://gmail.googleapis.com CALENDAR_API_BASE_URL = https://www.googleapis.com/calendar/v3 GITHUB_API_BASE_URL = https://api.github.com
✓ 4 API base URLs now point to Evalab twins
Compare outcomes
Let agents act, capture the full trace, and assess the behaviour with full transparency of the services' state.
Request activity · Outcome Passed
| Method | Request | Status | Latency |
|---|---|---|---|
| GET | /gmail/v1/users/me/messages/778 | 200 | 82ms |
| POST | /slack/api/chat.postMessage | 200 | 104ms |
| GET | /drive/v3/files/policy-9 | 200 | 68ms |
| POST | /github/check-runs | BLOCKED | 12ms |
The company
Everything your software needs to act, without the risk.
Twins
Full clones with API, MCP, and CLI support for third-party services your agents need to read from and write to.
Scenarios
Describe the scenario in natural language and get a seeded service with mock data, or seed the services with real data from prod.
Test runs
Automate evals after agents run in Evalab. Define judging criteria and grade performance for every commit or PR.
Evidence
Capture every provider call, response, latency, side effect, and service-state change across a run.
Places, please
Make your agents more capable in the real world.
Spin up isolated environments with API twins, full traces, and production-data for apps and agents that need to act.