Closed at 11pm. Handed off by morning.

Workflows, sequences, auto-fill and jobs do the busywork around every customer.

Workflows › Handover on WonCompleted · 1.9s

Whatever just happened can start a run.

Pick a trigger to see a run teams build on it.

Or add a “Run workflow” button to a view or dashboard.

Workflows › New deal setupOn
A new dealTrigger
Assign an owner
Add to the follow up list
Create the kickoff task

Record created fires when a record lands on the object you choose.

Read the order it runs in.

Steps and paths sit on one canvas, and every run keeps its log.

Give each path a condition

A fallback path runs when nothing else matches.

Try a step with a real record

Each step reads what the steps above it returned.

Loop over records

Run the same steps on every one.

Every run keeps a log

A failed run picks up from the step that failed.

Wait for the reply, not the calendar.

A sequence moves on when something happens, and stops for anyone who replies.

Wait forA durationA dateSomething to happenA task to be doneA call outcomeA callWhichever comes first

Records that keep themselves current.

Auto-fill looks after a field. A job works through every record that matches.

Auto-fill a field

On create, on update or on a schedule, in the field's own type.

Run a job

See the count first, then open any run to see what it touched.

Stella does the busywork, and asks first.

Stella answers from your records and can build a sequence for you. By default, you see every change before it's written.

  • Pick the model per agentGPT-4o, Claude, or Auto.
  • Web research is one toggleLeave it off and it won't search the web.
  • Focus areas steer itPoint it at the objects it works from.

Questions about automation.

What stops Stella changing data I didn't expect?

In Ask first, every write shows you the change: approve it, deny it, or always allow that kind (you can take it back in Settings). In Auto, Stella only stops before sending, deleting or archiving.

Can I limit what an agent looks at?

Give it focus areas. It prefers those objects and pages, and can still see the rest.

Can I pick the model?

Per agent: GPT-4o, a Claude model, or Auto.

What can start a run?

A record created or changed, a form, an email received or replied to, a conversation outcome, a call disposition, a chat on your website, or a “Run workflow” button.

How do I update records I already have?

Run auto-fill over a whole view, or use a job: filter records or conversations, check the count, then run it.

Can an agent reach the internet?

Only if you turn on web research for that agent.

Stop doing the part a computer should do.

Wire your first trigger in minutes.

Plans from $39 a seat a month, billed annually. See pricing