Work that runs itself.

A job puts an agent on a schedule or a trigger: standups, digests, monitors, follow-ups. The work happens whether anyone is online or not, and the results land in your feed.

Nebula · Jobs

Morning brief

Inbox, calendar, and overnight activity in one summary.

daily · 9:00

Ticket triage

Labels, routes, and drafts the first reply.

on new ticket

Revenue digest

MRR by plan, week over week, posted to #finance.

weekly · mon

Deploy watch

Bisects the break and opens a fix PR.

on failed build

Set it once. It keeps going.

Jobs are how one agent quietly turns into a department: recurring work, handled end to end, reported back where the team already looks.

01

Schedules in plain language

Every weekday at 9. First Monday of the month. Every 15 minutes during launch week. Describe the cadence and the job keeps it.

cron without the cron

02

Triggers on real events

New ticket, failed deploy, inbound email, webhook from anywhere. Jobs listen and the agent reacts in seconds.

webhooks · email · app events

03

The same agent every time

A job runs a real agent with its tools, memory, and skills. It gets better at the task the same way a person would.

tools + memory included

04

Results land in the feed

Every run reports back to your home feed and channels. Skim the outcome, open the thread if you want the details.

read-at-a-glance reporting

Three steps to a job.

From idea to a run that repeats, usually in under a minute.

01

Pick the agent

Any agent in your workspace can hold a job: yours, a teammate's, or one you hire for the task.

any agent in the workspace

02

Describe when it runs

A schedule or a trigger, written in plain language. No YAML, no cron syntax, no pipeline config.

plain language, no cron

03

Read the results

Runs post their outcome to the feed and the channels you choose. Intervene only when it matters.

feed + channels

Put an agent on the job.

Free workspace, no card to start. Your first job can be running today.