By the time you finish this walkthrough, your agent will be connected to your content, configured with clear instructions, and deployed to a real channel. Not a demo. Not a test environment. A working agent that real customers can use today.

Most agents that underperform were set up in a hurry. The content step got skipped. The instructions were left at their defaults. The agent got deployed before anyone tested it. This walkthrough is designed to prevent all of that. Each step builds on the last. Follow it once and you will not need to redo it.

In this article, you'll learn:

• What each onboarding step asks you to do
• What you can skip and come back to later
• How to get your agent live by the end

Overview of the Setup Flow

The onboarding flow has 5 steps:

1. Add your knowledge sources
2. Create your first action (optional)
3. Set up a handoff (optional)
4. Configure your agent
5. Customize your chatbot

A progress bar on the left side of the screen tracks where you are. You can skip steps 2 and 3 and return to them later - but completing all five gives you the strongest starting point.

Step 1: Add Your Knowledge Sources

This is the most important step. Your agent can't answer questions without content to learn from.

When you land on this step, Outlearn will already suggest a website crawl based on your email domain - so if your email is @yourcompany.com, it will suggest crawling yourcompany.com automatically.

From here you can:

• Accept the suggested crawl and let Outlearn index your website.
• Add Snippets - short written answers for specific FAQs.
• Add Knowledge - connect your Helpjuice KB, Google Drive, Confluence, or other integrations.

You don't need to connect everything now. Even one source is enough to get started. You can always add more from the Sources tab later.

When you're ready, click Continue.

Step 2: Create Your First Action (Optional)

Actions let your agent do things beyond answering questions - like checking a calendar, sending an email, or looking up a customer record.

If you want to set one up now:

1. Click + Create Action.
2. Follow the animated guide that walks you through the setup.

If you're not sure yet, click Skip for now - I'll set it up later. You can always add actions from the Actions tab.

Click Continue when done.

Step 3: Set Up a Handoff (Optional)

A handoff is what happens when a conversation needs to be passed to a real person on your team. You can connect Outlearn to Intercom, Zendesk, or email.

If you want to set one up now:

1. Click + Add Handoff.
2. Follow the animated guide.

If you want to skip it for now, click Skip for now - I'll set it up later. You can configure handoffs at any time from the Handoffs tab.

Click Continue when done.

Step 4: Configure Your Agent

This is where you give your agent its identity.

• Agent Name: What your agent is called - this appears in the chat widget header (e.g., "Helpjuice Support").
• Custom Instructions: These tell your agent who it is, how to behave, and how to handle specific situations. Outlearn pre-fills these based on your domain - review them and adjust them to match your tone and use case.

Good instructions make a big difference. Be specific about your company, your product, and how you want the agent to respond. You can update these at any time from the Settings tab.

Click Continue when done.

Step 5: Customize Your Chatbot

Before going live, you can personalize how the chat widget looks to your users.

• Chat Title: The name shown at the top of the chat widget.
• Welcome Message: The first message users see when they open the chat.
• Logo: Upload your company logo (PNG, JPG, or SVG).
• Primary Color: Set your brand color - a hex code works here.

A live preview on the right updates in real time as you make changes.

When you're happy with how it looks, click Go to Dashboard.

Step 6: Deploy Your Agent

Once you reach the dashboard, a prompt will guide you to deploy your agent.

1. Click Go to Deploy when the prompt appears.
2. On the Deploy page, click Customize Widget if you want to make further branding changes.
3. Click the Embed tab to get your embed code.
4. Copy the script snippet and paste it just before the </body> tag on any page where you want the chatbot to appear.
5. Click Done - your agent is now live.

If you're not ready to embed the widget yet, you can still test your agent using the Preview button in the top right corner of the dashboard - no embed code needed.

What's Next?

Once your agent is live, here's what we recommend doing first:

• Check your Sources - make sure everything synced correctly and toggle off any content you don't want the agent using yet.
• Test it with real questions - open the Preview panel and ask questions your customers actually ask.
• Review Analytics - after a few conversations, check the Content Gaps section to see what your agent couldn't answer and add that content to your Sources.

Best Practices

• Don't try to get everything perfect on day one. A working agent with one good source is more valuable than a perfect agent you never launch.
• The preview panel is your best friend - use it constantly to test changes before they go live.
• Revisit your Custom Instructions after your first week of real conversations - you'll quickly learn what needs adjusting.