Quick Start
Use this path to launch your first production-ready workflow.
Prerequisites
- Mimic workspace with admin access
- At least one agent provisioned (
rpaoropenclaw) - Test credentials for the target portal
- A sample CSV for batch validation
Step 1: Create your workspace baseline
- Invite operators and reviewers.
- Set plan and billing in
/app/settings/billing. - Confirm one organization admin can access
/app/admin.
Step 2: Provision an agent
- Open
/app/agents/new. - Choose mode:
rpa: deterministic browser automation from recorded steps.openclaw: conversational execution with tool calls.
- Wait until agent status is healthy before running tests.
Step 3: Record and generate workflow steps
- Upload a task recording in Studio.
- Verify generated steps and variables.
- Replace hard-coded credentials with secure variables.
Step 4: Run a single test job
- Launch a test from Studio or
POST /api/jobs/launch. - Confirm terminal status in Jobs.
- Inspect logs and screenshots for step-level quality.
Step 5: Move to batch mode
- Prepare a CSV with one row per case.
- Map CSV columns to workflow variables.
- Run a small canary batch (5 to 20 rows) before full volume.
Step 6: Monitor operations and ROI
- Track completion rate, flagged volume, and error taxonomy.
- Use org dashboards (
/app/org-dashboard/{orgKey}) to review latest batch funnel metrics, top blockers, and fix priorities. - Use queue analytics (
/app/org-dashboard/{orgKey}/analytics) to validate issue concentration by type, doctor, and office. - Review pilot dashboards for readiness and ROI trends.
- Export pilot evidence when required by partner review cycles.
Step 7: Operate the agent fleet
- Use
/app/agentsfor day-to-day controls. - Start or stop agents directly from the table when handling workload spikes.
- Rename agents in place to keep ownership and purpose clear for operators.