# First-launch checklist > A reusable release gate for agents, messages, automations, pages, campaigns, and connected channels. Canonical: https://help.spacebrain.ai/getting-started/first-launch/ Use this checklist before the first external launch and after any material change to identity, routing, content, audience, provider, or automation logic. ## 1. Define the outcome Write down the trigger, intended audience, expected result, owner, handoff, and evidence of success. “Send a follow-up” is incomplete; “send the approved follow-up email to a consenting test contact and create an owned task when they reply” is testable. ## 2. Validate identity and consent * Confirm the active workspace, sender identity, phone number, domain, and connected provider. * Confirm the audience source and the permission that allows each communication. * Include required sender identification, opt-out language, and regional disclosures. * Exclude suppressed, unsubscribed, invalid, internal-only, or otherwise ineligible recipients. ## 3. Validate content and behavior * Replace placeholder variables and verify fallback behavior for missing fields. * Check links, calendars, forms, tracking, and published destinations in a private browser session. * Test AI behavior with clear, ambiguous, adversarial, and out-of-scope requests. * Confirm the handoff path and what happens when a provider or action fails. * Review mobile layout and accessibility for public content. ## 4. Run a controlled test Use internal records and the exact production configuration where possible. Run once, wait for the result, and inspect the system of record before retrying. A second click can create a duplicate message, call, charge, registration, or workflow execution. Inspect the relevant evidence: * Inbox thread and contact activity for conversations. * Agent test result, call record, and recording for voice. * Workflow execution graph and node output for automations. * Form response, lead, published URL, and analytics for sites. * Task ownership and due date for operational handoffs. * Provider status and Spacebrain report for connected modules. ## 5. Limit the first audience Start with a small, representative group. Define the stop condition before launch: error rate, complaint, unexpected content, invalid routing, provider disconnection, or cost threshold. Keep a responsible operator available during the first run. ## 6. Monitor and document Record the launch time and timezone, configuration owner, audience size, expected completion window, and where failures will be reviewed. After the run, compare outcomes with the original success criteria and capture any required corrections. When an action appears stuck, check the destination and execution history first. Provider delivery can be delayed even after Spacebrain has accepted the action, and retrying too early can duplicate customer contact.