Creating Your First Chatbot

How to set up your first chatbot in 5 minutes.

To create a Bot, click on the “create a bot” button on the top right hand corner of the dashboard page.

Complete the 3 simple steps to create your first chatbot in 5 minutes.

Step 1: Filling in the Basic Details

  • Name of Chatbot – Use this empty field to name your chatbot as they appear to the customer

  • Tagline – Refers to the default tagline a user would see under the chatbot name

  • Chatbot color - Pick a color theme for your chat widget. The chatbot's header color will be the same as the hex code color chosen.

Step 2: Training your Chatbot on Data Sources

You can train your chatbot via multiple data sources, such as website links, files and videos as well as Google Drive folders

Types of Data Sources

Click below to expand 👇🏼

Website links 🔗

To train your chatbot on a website link, select the "URL" button. Insert the link to your website into the field.

If you wish to only crawl a specific page of your website, or a specific product folder, you can toggle the dropdown to change to the following options:

  • Entire website crawling - You can select the “entire website” dropdown to allow Wonderchat to store and access your whole website.

  • Subdirectory crawling - If you want to only store a subdirectory of your website, you can give a link such as “https://example.com/blog” and the tool will only crawl links with /blog/ at the back.

  • Single page crawling - If you want the chatbot to only crawl one page of your website, you can edit the dropdown to select “Single page”

Files and Videos (.pdf, .ppt, .docx, .csv, .json, .txt) 🗃️

Option 1: Uploading your own File & Video

You can train your chatbot on all file and video formats

To train your chatbot on these files, select the "File & videos" option and upload any file of your choice.

  • We accept PDF, CSV, TXT, DOC/DOCX, PPT, JSON formats for text files.

Option 2: Using Video Links

To train your chatbot on video files and youtube links, either upload any video file or add a Youtube link to a public video.

  • Youtube links provided must have a publicly viewable video link.

  • We accept video file formats of FLAC, MP3, MP4, MPEG, MPGA, M4A, OGG, WAV, WEBM (Audio/Video).

Google Drive files 📁

To train your chatbot on Google Drive files, click on "Connect to Google Drive" to train it on the drive's content.

URLs to Ignore When Indexing

To make sure that your Chatbot does store old links, we would recommend that you ignore URLs that you do not want the chatbot to index.

💡 The difference between ignoring URLs and deleting URLs is that the former will completely exclude those URLs from ever being crawled. This means that if you choose to reindex your chatbot, ignored URLs will not be crawled.

Deleting urls is a one off deletion, and deleted URLs are removed from your chatbot. However, in the event of a re-crawl, webpages that were previously deleted would still reappear.

How to input URLs to ignore from your site when indexing

To ignore a single page url, copy and paste the url of the specific page you wish to ignore.

To ignore an entire subdirectory of pages such as https://wonderchat.io/blogs you will need to add an asterisk (*) behind the url name.

Once the link is ignored, it will not be recrawled for your monthly chatbot recrawl or if you press the recrawl button.

Step 3: Customize your chatbot's responses

Selecting the best AI model for your chatbot

We offer a diverse selection of 10 AI models, each excelling in different areas. Our lineup includes models from:

  • OpenAI

  • Claude

  • Gemini

  • Llama

  • Minstral

  • Cohere

  • and more.

The most popular models that have been proven to function well are the Claude Sonnet models and the GPT-4 Turbo models.

For a more detailed overview of which is the best chatbot model to pick from, you can view the AI model comparison and analysis here.

Selecting Between AI models for your chatbots

Defining the role of the chatbot

To customize your chatbot's responses, decide on the role of your Chatbot. Users can either choose a pre-made role prompt for their chatbot, or write their own custom prompt.

Pre-made Role Prompts

Pre-made role prompts give your chatbot a personality and default brand voice according to the role selected.

  • Copywriter - A chatbot set to the role of a copywriter will behave like a marketing copywriting assistant.

  • Salesperson - A chatbot set to the role of a copywriter will behave like a sales person and try to sell your website to your website visitors.

  • Customer Support agent - A chatbot set to the role of a customer support agent will behave like a customer support staff and try to help your website visitor troubleshoot their issues.

  • Researcher - A chatbot set to the role of a researcher would be able to act like a researcher and make references to information outside of the webpages stored within their knowledge base.

  • General - A chatbot set to the role of a general chatbot would be able to answer any questions limited to the scope of your knowledge base.

Custom Prompts

Custom prompts allow you to customize the role of the chatbot according to the brand personality and voice that you desire.

To set your own custom prompt, select the "Write your own prompt" option and edit the prompt field to configure the role of your chatbot.

The custom prompt helps you define your own specific instructions for your chatbot to follow.

You can use the following sample prompt as a template to write your own custom prompt:

Sample Prompt

You are [Name of Chatbot], an AI [Role of Chatbot] at [Name of Company]. Embodying the traits of helpfulness, professionalism, cleverness, and friendliness, your primary role is to assist users with their queries regarding [Name of Company]. You act as an expert in all things [Name of Company], providing answers and assistance based on your extensive knowledge of the product.

Remember, your responses should be based exclusively on your understanding of [Name of Company]. Abstain from responding to queries unrelated to [Name of Company], and always adhere strictly to the following rules:

RULES:

1. Always adhere to these rules without exception.
2. Your answers should solely be based on your comprehensive understanding of [Name of Company]
3. Do not respond to queries unrelated to [Name of Company]
4. If a user asks a question unrelated to [Name of Company], respond with "I apologize for any inconvenience, but I’m not sure about this. For additional assistance, please reach out to our support@yourcompany.com.
5. Only share links that are part of your source data or the official [Name of Company] sources.
6. If you cannot find an answer within your [Name of Company] knowledge base, guide users to the support page at [Link to support page] so they can contact [Name of Company] Support.
7. Consistently maintain your role as [Name of Chatbot], the [Name of Company] AI [Role of Chatbot], in all interactions.
8. Ensure that every answer you provide is something you are confident is true, based on your [Name of Company] knowledge.

By strictly adhering to these rules, you will effectively fulfil your role as [Name of AI], providing users with accurate and trustworthy information about [Name of Company].

Setting the Chatbot's Default Language

You can set the default chatbot language by selecting the language dropdown below.

  • English is the universal language, a chatbot can switch between English an any language when responding to user queries.

  • Setting a non-english language will force the chatbot to default to the non-english language only.

Greeting and Fallback Messages

The greeting message field allows you to set custom responses to greet website users, and the fallback message fields allows you to answer them in situations where the chatbot is unsure of what to respond.

Greeting Message

The greeting message field allows you to set custom responses to greet website users at the start of each chat.

Fallback Message

Similarly, the fallback message appears when the chatbot is unable to respond to a user query. Customize this message to fit your brand voice.

Once finished, you can click on ‘create’ and your chatbot will load-up.

The average bot that scans a website with 100 pages will load in around 5 minutes. Bigger websites can expect a longer loading time, but you will receive an email in your inbox once your chatbot has completed its training.

If you have any more questions, feel free to reach out to us at support@wonderchat.io

PreviousSetting up your chatbotNextHow to set-up a Spontaneous Greeting Message

Last updated