> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getnexor.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Launch checklist

Every Nexor onboarding closes by validating three things. If all three are true, you are ready to activate or to keep iterating on your own. If one is not, the linked page fixes it.

## The three validations

<Steps>
  <Step title="You understand what an agent is">
    You can name the blocks of your agent and say which one to change when it misbehaves: goal, variables, identity, business rules, knowledge base, pipeline, contact settings, channels, tools. If not, reread [Anatomy of an agent](/docs/en/guides/getting-started/anatomy-of-an-agent).
  </Step>

  <Step title="You know how to test it">
    You have run your own scenarios in the playground, made a change from NexorGPT, trained, and seen the change take effect. If not, go to [Test in the playground](/docs/en/guides/getting-started/test-in-playground).
  </Step>

  <Step title="You know how to connect or advance your channels">
    You know which channel opens the conversation, it is connected or its connection is in progress, and you know how leads will get in. If not, go to [Operation and channels](/docs/en/guides/getting-started/operate-and-connect-channels).
  </Step>
</Steps>

## Before you activate

* **Goal** matches the final state from your worksheet.
* **Required variables** cover everything you need to qualify or complete the goal.
* **Business rules** include limits, escalation triggers, and discard rules, and the agent was trained after the last edit.
* **Knowledge base** sources are enabled on the agent.
* **Pipeline** stages mirror your process, and the winning stage is the one you expect.
* **Contact hours and re-engagement attempts** match how a good seller on your team would behave.
* **First-contact channel** is connected, and templates are approved if it is WhatsApp.
* **Leads** are ready to import, starting with a small batch.
* **Inbox** ownership is clear: someone on the team will read conversations and take escalations during the first days.

## After you activate

* Read the first conversations in the Inbox and note anything you would change.
* Ask NexorGPT "what can I improve in my agent?" and, for any odd reply, attach the lead and ask why. See [Debug a lead](/docs/en/guides/assistant/debugging-leads).
* Adjust one thing at a time, train, and compare.
* When the flow is stable, scale the lead volume and add channels.

<Tip>
  Finish operating, or with a base agent ready to iterate. Do not finish with an open list of pending items. If something blocks you, resolve it now with the linked guide or with NexorGPT.
</Tip>

## Where to go from here

<CardGroup cols={2}>
  <Card title="Prompt best practices" icon="list-checks" href="/docs/en/guides/agents/prompt-best-practices">
    Sharpen each section of the business rules.
  </Card>

  <Card title="Connect WhatsApp" icon="message-circle" href="/docs/en/guides/connecting-whatsapp">
    Link your business number and get templates approved.
  </Card>

  <Card title="Connected agents" icon="git-branch" href="/docs/en/guides/getting-started/connected-agents">
    Split a process into agents with transfers.
  </Card>

  <Card title="Advanced automation" icon="workflow" href="/docs/en/guides/advanced/overview">
    Webhooks, tools, jobs, and functions.
  </Card>
</CardGroup>