Blogs > Latest > How to build your first AI Chatbot

How to build your first AI Chatbot

Syed Raza's profile picture
Syed Raza

There are a few steps to setting up an AI Chatbot. First, we must Provision an AI Chatbot. Second, we must create a database with all our relevant content. Third, we connect the database to our AI Chatbot adjusting similarity score. Lastly, we embed the AI Chatbot on our website! In this blog we will learn to:

  1. Provision an AI Chatbot ๐Ÿค–

    Creating an AI Chatbot on Splutter AI - the fastest way we can! Add a name, profile picture & thats it!

  2. Creating a Database ๐Ÿ›ข๏ธ

    Uploading neccessary content to your database - use Files, QA Pairs, Text, Web Pages, anything!

  3. Connecting the Database to the AI Chatbot ๐Ÿ”—

    Making the database & AI Chatbot work together - the most important step! Databases are fully hot-swappable on Splutter AI!

  4. Testing our AI Chatbot via Prompting ๐Ÿงช

    Lets test our AI Chatbot in the "Behaviour" Tab where we can monitor our AI Chatbots performance & make changes when needed!

  5. Embedding our AI Chatbot on a Website ๐ŸŒ

    Following the steps listed out in the "Embed on Site" Tab on our AI Chatbot Dashboard, have our AI Chatbot embedded on a website!

  6. Monitor Chat History ๐Ÿ–ฅ๏ธ

    Now that our AI Chatbot is live, we can monitor the chat history to see how our AI chatbot is performing & make changes when needed!

Vector Database

Provision an AI Chatbot ๐Ÿค–

  1. Sign in to Splutter AI

    Start by signing in to your Splutter AI account.

  2. Provision an AI Chatbot: Choose a Model

    Choose a Model
    Select the desired configuration and settings for your AI Chatbot.
  3. Provision an AI Chatbot: Name AI Chatbot & Upload PFP

    Name AI Chatbot & Upload PFP
    Name your AI Chatbot and upload a profile picture.
  4. Use AI Chatbot: Go to Embed Tab

    Embed Tab of AI Chatbot
    Open your AI Chatbot & go to the Embed Tab
  5. Use AI Chatbot: Make Chatbot Public

    Make AI Chatbot Public
    Make our AI Chatbot public so you can debug/prompt engineer/for others to interact with.

Great! Now that we have our AI Chatbot ready, we need to create a database & connect it to our AI Chatbot. The databases use retrieval augmented generation - The idea is that when a user sends a query/message to our AI Chatbot, we will first send that query to the database to check & fetch all relevant documents - the AI model will automatically add the relevant documents & use that to respond to customer queries! Lets do that now!

Creating & Adding Content to our Database

Creating a Database ๐Ÿ›ข๏ธ

  1. Setup Database Provisioning

    In the Dashboard click the "+" Icon beside Databases, this will bring you to the menu below. You can name your database & choose the default embedding option

  2. Step 1: Provision New Database

    Access Database Page
    You can name your database & choose the default embedding option. Once this is done, the page will refresh & you will see your new database! Go ahead & click it
  3. Step 2: Navigate to Database Page

    Configure Database Settings
    Click your database once provisioned & you select it, it will take you to https://www.splutter.ai/databases where you can select & edit your database - Choose your database here to get started adding content

Add Content to Database ๐Ÿ“

  1. Adding Various Types of Content

    Here you will add different types of content to your database - Files, QA Pairs, Text, Web Pages, anything! - The free plan allows up to 10 links however you can add more content via Text or QA Pairs or Files if needed. The free plan allows a generous 50,000 Characters per database

  2. Initial Look: Containers

    Add Containers
    Add containers to organize your content. This is only for your personal use & does not affect database performance or abilities - You can segregate API Documentation, Customer Support, Sales Information over here with color coding
  3. Step 1: Add Text Content

    Add Text Content
    Add textual content to your database. This content will automatically get chunked into chunks called "vectors" but its just a chunk of text. You can fully edit each chunk
  4. Step 2: Add Web Content

    Add Web Content
    Add content from web pages to your database by adding a URL & scraping (it will also fetch related links) - Each web scrape costs 1 query token. Each scrape also fetches related URLs on the same page to help you with your scraping! You can "Scrape All" or you can disable the blue toggle & add the URL without scraping & manually add content to the chunk
  5. Step 3: Add File Content

    Add File Content
    Upload files to add their content to your database & fully edit every chunk of text added. Usually this is good for PDfs, DOCs, TXT, Simple Code Files & other text based formats
  6. Step 4: Add Q&A Content

    Add Q&A Content
    Add question and answer pairs to your database & add more when needed! Usually when looking at our AI Chatbot chat history, you will find unique scenarios that need to be accounted for & this is the place to deal with them!
  7. Step 5: Update Database *IMPORTANT*

    *Update Database*
    Update the database with all changes it will upload & embed all the content for your AI Chatbot

Great! Now that we have our Database is ready, we now need to connect our database to our AI Chatbot! This is really easy to do & Splutter AI allows for hotswappable databases so lets get started!

Connecting the Database to our AI Chatbot ๐Ÿ”—

  1. Enabling and Configuring Database

    Connecting the Database to our AI Chatbot - This is really easy to do & Splutter AI allows for hotswappable databases so lets get started!

  2. Step 1: Open AI Chatbot & Go to Database Tab

    Navigate to Database Tab
    Navigate to Database Tab for Specific AI Chatbot
  3. Step 2: Enable Database

    Enable Database
    Enable the provisioned database for your AI Chatbot.
  4. Step 3: Modify Similarity Settings

    Modify Similarity Settings
    Adjust similarity settings to fine-tune the database query results. Higher the number, the more likely results will be added to context for the chat
  5. Step 4: Configure Items/Query

    Configure Items/Query
    Configure the items and query settings for optimal database performance. More items should only be done for AI Chatbots that can handle high context length
  6. Step 5: Click Update on the right

    After clicking the Update button on the right, the database selection will be updated & the database will be officially connected to our AI Chatbot!

Great! Now that our AI Chatbot-Database connection is complete, we can now test our AI Chatbot via Prompting! This is really easy to do on Splutter AI we simply open the "Behaviour" Tab on our AI Chatbot Dashboard where we can chat with our AI Chatbot & see how it is performing!

Testing our AI Chatbot via Prompting ๐Ÿงช

  1. System Instruction of AI Chatbots

    AI Chatbots are guided by a system instruction, which sets the overall behavior and tone of the AI Chatbot. This instruction acts as a high-level directive that the model follows, influencing how it interprets and responds to user inputs. Splutter AI offers a prompt engineering dashbaord where users can systematically modify their prompt, critique it, then update their prompt for their AI Chatbots.

  2. Behavior Tab

    Behavior Tab
    Modify the system instruction in the "edit" tab by clicking role, instructions, objectives, tools, script, examples, policies, & more!
  3. Behavior Tab

    Behavior Tab
    Here is the "view" tab to critique it which will give a rating to your prompt, lastly click: "update" to update the chatbot with your new prompt
  4. Chat with our AI Chatbot in the "Behaviour" Tab

    On the left is our live chatbot that we can chat with & see how it is performing! If you want faster responses, you can go into Chatbot Settings & Enable Streaming to get real-time responses!

Excellent! We are almost done! Now that our AI Chatbot has been tested & is performing well, we can now embed our AI Chatbot on our website! This is really easy to do on Splutter AI we simply open the "Embed on Site" Tab on our AI Chatbot Dashboard where we can embed our AI Chatbot on our website!

Embedding our AI Chatbot on a Website ๐ŸŒ

  1. Embed Chatbot: Copy the Embed Code

    Copy the Embed Code
    Copy the embed code for integrating our AI Chatbt - Here we can choose which technology a website is built on & follow the steps to embed our AI Chatbot!
  2. Embed Chatbot: Example React

    Example React Embed
    Example of embedding the chatbot in a React application
  3. Paste in HTML or React Document

    Insert the embed code into your HTML or React document to embed the chatbot.

  4. Share URL

    Alternatively, you can go to the share tab & try our AI Chatbot our live & even share the URL with your friends!

  5. Paste URL in iframe

    You can also directly embed using an iframe by pasting the URL in an iframe - even using our AI Chatbot as a Bot Funnel (instead of a Website Funnel!)

Magnificent! Now that we have successfully embedded our AI Chatbot on our website, we can now monitor the chat history to see how our AI Chatbot is performing & make changes when needed! This is really easy to do on Splutter AI we simply open the "Chat History" Tab on our AI Chatbot Dashboard where we can monitor the chat history!

Monitoring Chat History ๐Ÿ–ฅ๏ธ

Chats Overview

The "Chats" section displays a list of all chat sessions, ordered by the most recent message. Click on any chat to fetch and view the entire conversation. This allows you to monitor and review the interactions your users have had with your chatbot.

Navigate to Chats Tab
Navigate to the Chats Tab on your Chatbot Navbar
Chat History Screenshot
This image shows the chat history interface where you can see all the interactions in chronological order.

Tickets Overview - [Ticketing Tool Required!]

The "Tickets" section lets you manage support tickets that have been submitted via our AI Chatbot. Remember this requires buying the "Hobby Plan", setting up a Ticket Tool & adding it to our AI Chatbot! Each ticket can be updated with additional messages, and once resolved, can be marked as closed. This helps you keep track of ongoing issues and ensures timely resolution.

Tickets Management Screenshot
This image illustrates the ticket management interface where you can add messages or close tickets. Email is a required field for tickets!

Leads Overview - [Leads Tool Required!]

The "Leads" section provides a list of potential customers or users who have shown interest in your services. Remember this requires buying the "Hobby Plan", setting up a Leads Tool & adding it to our AI Chatbot! You can view detailed information about each lead and download the leads in CSV format (if leads exist) for easy integration with other tools like CRM systems.

Leads Management Screenshot
This image shows the leads management interface where you can view and download captured leads.

Conclusion

Congratulations! ๐ŸŽ‰ We have succesfully setup our first AI Chatbot from scratch! In this blog we provisioned an AI Chatbot ๐Ÿค–, created a Database ๐Ÿ›ข๏ธ, populated our Database with relevant data ๐Ÿ“, connected our AI Chatbot to our new Database ๐Ÿ”—, tested our AI Chatbot ๐Ÿงช, embedded our AI Chatbot on a website ๐ŸŒ, & learned how to monitor our AI Chatbots activities ๐Ÿ–ฅ๏ธ - It doesn't get better than that! Now that you have everything you need to know lets get building! ๐Ÿ› ๏ธ

Only Need Query Token? Just take what you need!

Need a Plan? Try our Hobby Plan!

Powered by Splutter AI