Automations | OrbiLattice

OrbiLattice Help Center

Beta

Workflow automation

Build, test, and operate versioned workflows in your workspace. The capability catalog and node inspectors show which configurations and executors are currently available.

Open builder Workflow operations

Choose a starter or prepare an AI draft

The starter library includes follow-up checklists, scheduled review tasks, deal-stage diligence, and a rental-return decision. Needs setup means you must supply inputs, select records, or connect an eligible service. Unavailable templates are preserved legacy designs and cannot be applied.

Use the builder’s AI button to describe a workflow with concrete inputs and outcomes. Prepare AI draft opens a review before replacing your canvas. Check the steps and setup requirements, then choose Use this draft. Applying a template or AI draft leaves it unsaved and inactive. If AI is unavailable, your existing graph stays in place. The floating app assistant does not currently create or save native workflow graphs.

1. Author an inactive draftOpen the workflow builder. Start with a manual trigger and an available calculator or internal action. Connect nodes, configure their inputs, resolve validation issues, and Save. Saving does not activate the workflow.
2. Preview and test without effectsStructural Preview checks the graph without recording an execution. After saving, select the manual trigger and use Start Test Run for a persisted simulation. Test results do not prove that a live provider or action will succeed.
3. Publish an exact versionPublish a validated graph. The published version is immutable and identified by its checksum. Publishing alone does not activate it.
4. Activate explicitlySelect Activate after publishing. The server checks the published version and current workspace permissions, Workflow Automation access, and capability readiness before enabling live operation.
5. Confirm a live runIn the manual-trigger inspector, choose Review Live Run, inspect the configured actions in the graph, then Confirm Live Run. Live actions can change records or contact external services. Automatic runs require a supported trigger registered against the published version.
6. Inspect and recoverOpen History or Workflow operations to inspect durable runs, node attempts, errors, and sanitized delivery evidence. Pause, cancel, resume, retry, and repair appear according to your permissions and the current execution state.

Read the dashboard

Executions, Success rate, Failed runs, and Unknown effects summarize recorded runs for the selected 24h, 7d, or 30d window. Select a workflow to inspect its execution history and traces. Refresh retrieves new evidence; Dashboard appearance changes the theme. These metrics do not measure revenue, ROI, or AI accuracy.

An unknown effect means an action may have happened but its outcome is unresolved. Inspect the evidence before another attempt. Repair requires confirmation of whether the specific effect was applied.

Email, SMS, webhooks, and AI actions require their own workspace readiness, consent, provider, and access checks. Supported schedules use the UTC configuration shown in the inspector. An available node or a successful simulation does not establish live delivery. Follow the native controls; this guide does not authorize or execute a run.