Skip to main content
In just a few minutes, your agent will start answering questions about your business and engaging with your visitors. Let’s begin.

Prerequisites

You’ll need an active Watermelon account. If you don’t have one yet, sign up first.

What You’ll Do

1

Create your agent & personality

Set up a new agent and write a personality that fits your companies brand guidelines.
2

Add sources

Connect and manage the data sources your agent will use to learn from — such as your website, files, or the world wide web.
3

Test & optimize

Test answers in the tester and fine-tune behavior and responses.
4

Publish & go live

Add your agent to your website or connect it to your favorite channels.
Estimated time: 5 minutes from start to finish

Step 1: Create your agent & personality

After signing into your Watermelon account, go to the agents overview. Click the “Create new AI Agent” button to get started, provide it with a name and click continue.
Agents Overview Pn

Write your personality

After you created your agent, you will be in the domain knowledge section where you can provide a personality. After writing the personality click save.
Personality Pn
Tips for your agent’s PersonalityYour agent’s personality defines how it sounds and behaves in conversations. A good setup helps it feel natural, trustworthy, and aligned with your brand.What to include:
  • Tone of voice: Should it sound friendly, professional, or neutral?
  • Communication style: Short and clear, or more conversational?
  • Formality: Use “you” or “we”? Keep language consistent with your website tone.
  • Role in your organization: Describe how the agent represents your company — e.g. customer support assistant, sales guide, or digital host.
  • Behavior guidelines: How should it respond when unsure, when users are frustrated, or when a question is out of scope?
Start small — define 3–4 personality traits (e.g.calm, human, clear ) and test how they sound in real conversations. Then adjust based on feedback.

Step 2: Add sources

Your AI agent needs information to learn from. You can train it using various data sources:
  • XML feed
  • Web Crawler
  • Doc Scraper
Add your product XML feedImport structured data directly from your feed to keep your Agent up to date automatically.Best for: Product catalogs, listings, or frequently changing information
When you upload a document, its content is added to your Agent’s knowledge almost instantly. Stay on the page until the document has been fully processed.
Xml Feed Pn
Limits: Different plans have different upload limits for training data. Check your plan if you hit any limits.

Step 3: Test & optimize

Access the tester

Once you’ve added all the relevant information, you’ll can access the interactive tester - your testing environment where you can chat with your AI Agent and fine-tune it before publishing it.
Playground Pn

Evaluate Response Quality

As you test, look for:
Accuracy - Are responses factually correct?
Relevance - Does it answer what was asked?
Completeness - Are responses comprehensive but concise?
Tone - Does it match your brand voice?

Fine-tune settings

  • AI Model
  • Instructions

Test responses with different models

1

Change the model

You can change the agent’s model in the agent profile.
Agent Profile Pn
Model Change Pn
2

Test with different models

Ask the same questions to different models and compare:
  • Response quality and accuracy
  • Response time and speed
  • Tone and personality
  • Handling of edge cases
3

Make your decision

Based on the comparison, select the model that best fits your specific use case and brand voice.
For detailed model comparisons and advanced testing strategies, check out our comprehensive model comparison guide.

Step 4: Publish & go live

Once you’re satisfied with your AI Agent’s responses, it’s time to make it live! Navigate to the “Activate” tab in your agent, and click on “Connect channel” button to connect your desired channel.

Activate agent

After you have connected your channel, click ‘activate’ and your agent is now live on the connected channel!

🎉 Congratulations!

You’ve successfully created, tested, and published your first AI Agent! Here’s what you’ve accomplished:
Created an intelligent AI Agent trained on your data
Tested and validated response quality
Deployed it live on your website
Made it accessible to your visitors 24/7

What’s next?

Need help? Our support workspace is here to assist you. Contact support or check out the FAQ section for common questions.