Getting started

Estimated reading: 5 minutes

Welcome to Kaila, a platform for creating your own generative AI-powered agent for natural language understanding and conversation. To get started, you’ll need to create a Kaila Account and log in to our platform. Once you have access to your account, you’ll be able to create your own Kaila Agent.

This quickstart guide is designed to get you up and running with a Kaila Studio in a snap. Along the way, you will better understand what Kaila is, setup your Kaila Account, get a hands-on introduction to the Studio and Integrations, and find more resources to deep-dive into.

1. Creating an Account and Logging In:

The first step in using our platform is to create an account. To do this, visit our website and click on the “Early Access” button. You will be prompted to enter your name, including your surname, email address, and to create a password. Once your account has been created, you can log in by entering your workspace URL (something like your-company.kaila.ai) or easily access the Kaila.ai website and click on the “Login” button provided in the top right corner. Logging in through our website, Kaila.ai, is the easiest form of login because you don’t need to remember your workspace URL, thanks to the “Workspace finder.” When you get to the login page, please provide your email address and password in the designated fields.

2. Selecting and Preparing Your Dataset:

After logging in, you will be taken to your Kaila Studio dashboard. Here, you can select the knowledge data you would like to use to train your Kaila Agent. At this time, data can only be in the form of plain text. Before starting the training process for your Kaila Agent, please refer to the article “How to Prepare Your Training Data.“

Please keep in mind:

“Preparation of your dataset and proper selection of the type of document is essential and crucial when creating a Kaila Agent. Kaila is a powerful generative AI, but there is one basic rule to keep in mind: ‘trash in, trash out.’ It is important to provide Kaila with high-quality data to ensure the best results.”

3. Create Your First Generative AI-powered Agent

Once your knowledge data is prepared, you can use Kaila Studio to train your generative AI-powered agent. The platform offers a user-friendly interface that allows you to build and customize your agent. You can train your agent to understand specific topics and respond to user inquiries in a manner that is tailored to meet your business requirements. Upon completion of the training, you can integrate your agent into various channels, such as websites, mobile applications, and messaging platforms, to provide your customers with a personalized and efficient conversational experience

Beta Limitation: Thanks to the high demand of beta users, we have set the maximum size of the dataset to 15,000 words (roughly 5 characters per word). But don't worry - after we release the first stable version, our users will be able to connect to any size of data sources or upload any size of files.

4. Test & Evaluate

Here comes the most exciting moment. Test and evaluate your new Kaila Agent with a simple chat interface where you can ask questions not only about the topics and information you have trained it on, but also test its behavior in limiting situations. For example, you can ask how many legs a horse has or what color the sun is. Don’t worry about asking sexual or offensive questions, as Kaila has built-in intelligent filters to handle such cases and will not answer them.

If Kaila is not answering your questions correctly or the answers are misleading, the problem may not be that the generative model is not smart (although it is still a machine and has its limits, sometimes resulting in hallucinations like all generative models), but rather in the dataset itself. Kaila may lack the necessary data, subject matter, or context to answer the question accurately. Check out our article about dataset preparing best practices.