Product / Routines
Make useful work repeatable.
Run agent tasks on a schedule, on demand or through webhooks, with visible results.
Write a routine with a finish line
A routine needs a task, inputs and a definition of success. For a weekly brief, name the folder to read, the period to cover, the output file to create, and what must wait for your review. Run it manually before adding a schedule.
Schedules and waking
Hosted preview checks have exercised a scheduled run and waking a sleeping sandbox before a due routine. This is not a promise that every unattended workflow is qualified. Long-running turns across idle transitions still need further verification.
Use webhooks with retries
A webhook can trigger a task from another service. When the sandbox is asleep, the first request can receive a temporary 503 response while it wakes. Configure sender retries; webhook deliveries are not currently queued for you. Check the run record before treating a trigger as a completed task.
Keep exploring
Your first agent, step by step.
Set up the Suparagent hosted preview with an email address, your sandbox-provider key and your AI account.
Security & dataUnderstand where the work runs.
Learn the boundaries between your browser, the hosted service, sandbox providers and AI engines.
Use casesPut a team around a real task.
Explore practical agent workflows for research, development, content and operations.
Give the work a home
Build your first AI team.
Start with one agent, one clear task, and a result you can review.
Preview · Bring your sandbox provider and AI account