Beunec

Platform

Automation

Automation is the Aselius app for scheduled engine runs. A run happens in the cloud on your schedule, whether or not your computer is on, and every result is checked independently before it is marked as succeeded.

At a glance

What it is
scheduled, unattended runs of a Rivine engine, part of Aselius.
Engines
Rivine Research, TaskForce, Cicero and Confluence. Browse, Code and Assist cannot be scheduled.
Safety
reads public sources only, never acts on your accounts, and starts only after a test run passes.
You get
a fresh taskspace per run, marked Succeeded or Needs review, with a notification.
Cost
billed to your credits at each engine's usual start floor.
Where
open Automation from the platform rail at aselius.beunec.com.

In the product

Real screenshots of what this page describes. Open any of them full size.

Which engines can be scheduled

EngineCan be scheduledWhat an unattended run does
Rivine ResearchYesObserves, researches and saves cited briefs
Rivine TaskForceYesBriefings and structured deliverables
Rivine CiceroYesRegulatory watch and drafts for your review. It never files or signs anything.
Rivine ConfluenceYesRefreshes forecasts and scenarios
Rivine BrowseNoNeeds your own browser and signed-in sessions on your device
Rivine CodeNoUseful scheduled work would need write access to your repository
Rivine AssistNoEdits the document on your screen, so it is interactive by nature

Setting up an automation

Each automation has an engine, a goal, a budget and a schedule. You can start from a template or write your own.

  1. Create it as a draft. Choose the engine, describe the goal, set a budget and a schedule (recurring or one time).
  2. Run a test. The test run uses exactly the engine, goal and budget you set.
  3. Activate it. An automation becomes active only after a test of this exact setup passes, or after you review and approve a test result that was held for review.
  4. Let it run. Scheduled runs start at each time you set.

Note: Changing the engine, goal or budget returns the automation to draft, and it needs a new passing test. Changing only the schedule does not.

How a run happens

  1. Checks before starting. The automation must be active and unchanged since its passing test, and your balance must meet the engine's start floor.
  2. A fresh taskspace. Each run starts a new engine turn in its own taskspace, with a short summary of the previous result for continuity.
  3. No connectors. A run can read public sources and create work in your workspace. It never connects to or acts on your accounts or connected apps.
  4. Independent check. Automatic checks run first, then an independent review that sees the goal and the result but not the engine's reasoning.
  5. Result and notification. A run that passes is marked Succeeded; otherwise it is marked Needs review. You are notified either way, and the run's status, timing, cost and taskspace are kept in the automation's history.

Retries and pausing

SituationWhat happens
Temporary errorRetried twice, a few minutes apart
Out of creditsThe run stops and the automation pauses
Refused by policyThe run stops and the automation pauses
Three consecutive failuresThe automation pauses

You can pause an automation yourself at any time, and edit or delete it.

Lifecycle

StateMeaning
DraftBeing set up or edited. Does not run on its schedule.
ActivePassed a test of its current setup. Runs on schedule.
PausedStopped by you, by a credit or policy stop, or after three consecutive failures.
CompletedA one-time automation that has run.

Billing

Each run is billed to your prepaid credits like any other engine turn, at the same per-engine start floor as the web app. See Plans & Pricing.

Related