antonagents

sign in to your workspace

antonagents

Delegate a job

Say what you want done. Get the outcome.

Describe the job and when it should happen. You review the plan — instructions, schedule and the checks that prove each run really happened — then it runs, recovers on its own, and only asks you when it has to.

or work it out in a chat first, then save it as a job
Or set up a routine manually
What to do — model, connectors, skills, data & secrets
For databases, use Data sources in the sidebar (encrypted) — don't paste connection strings here.
+ New skill

Your routines

Select a routine to see its runs.

Outcomes

Jobs delegated to your routines: how many were verified done, how many needed nobody, and what each one cost.

Connectors

Connect your accounts once — every chat and routine can then act through them, scoped to exactly what you grant.

Data sources

Connect a database once; attach it to a routine, or just ask about it in chat. Connection strings are encrypted at rest and never shown again.

Connect a database

Read-only is only guaranteed when the login itself can't write. We check the login's permissions when you test or save, and show you how to fix it if it can write.

Alerts

Notifications from your routines — failures, matches, and missed scheduled runs.

Members

People in this workspace.

Observability

Oversight of your agents, org-wide: health, cost, and a full activity audit of what every agent did.

Activity audit