KChat

KChat is our AI chatbot solution.

To use KChat, you will need Kurator. If you have already registered with Kurator, you can use your credentials to log in here.

If you want your KChat registration to be different from your Kurator registration, you can register and authenticate your account on the KChat website.

Registering KChat

If you choose to register your KChat account separately from your Kurator account, visit the KChat app at https://kchat.website and register a new account.

After registering, you will receive an email to authenticate your account. Once authenticated, you are ready to log in.

Buying A Plan

Before you can get started with KChat, you will need to buy a plan.

Currently, we do not offer any free trials. If you want to test KChat, you can buy a basic plan and unsubscribe if you wish to discontinue.

After purchasing a plan, you will get access to the dashboard. You can then upload your knowledge base and create your chatbot.

Upload Your Knowledge Base

Your first action, for creating your chatbot is uploading your knowledge base.

A knowledge base in KChat is a file you create and manage in Kurator.

The files are created under the "WordPress KBucket Files" of your Kurator app.

Here, you can create a file, and attach one or more folders you like to use as a knowledge source for your chatbot responses.

Knowledge Base Demo

Adding KChat

After you add your knowledge base to your KChat account, you are ready to create your KChat chatbot.

Here are the steps for creating your chatbot:

Give the Chatbot a Name
Select the temperature for your chatbot
Select the LLM to be used with your chatbot
Select the Knowledge Base
Select the outbound links

Chatbot Tempreature

The "temperature" in the context of a language model (LLM) like GPT-4 refers to a parameter that controls the randomness of the model's output. It essentially determines how confident or adventurous the model should be when generating responses. The temperature setting affects the distribution of possible next words in the sequence, impacting the diversity and creativity of the generated text.

Here’s how it works:

Temperature Parameter: The temperature is a floating-point value, typically between 0 and 1, but it can also be greater than 1.
- Lower values (e.g., 0.1): Make the model more conservative. It will produce more deterministic and focused responses. This is useful for tasks where accuracy and coherence are critical.
- Higher values (e.g., 1.0): Make the model's output more random and diverse. This can be useful for creative tasks where variety and novelty are desired.
Probability Distribution: The temperature parameter modifies the probability distribution of the next word.
- Low Temperature: The model heavily weights the most probable words. This leads to safer and more predictable responses.
- High Temperature: The model flattens the probability distribution, giving less probable words a higher chance of being selected. This results in more varied and unexpected responses.
Practical Use Cases:
- Low Temperature: Suitable for applications like customer support, where consistency and reliability are crucial.
- High Temperature: Ideal for creative writing, brainstorming, and scenarios where unique and varied outputs are beneficial.

Chatbot LLM

In the drop-down menu for your LLM, you will currently find only one option, GPT 3.5.

You can see a list of other LLM's greyed out. In the near future we will be offering you more options, including a custom LLM that will cost less and do more!

Attach Knowledge Store

In this section, you will see a list of Knowledge store you have added to your KChat app. You can choose one or more knowledge stores as the source for your chatbot responses.

Outbound Reference Links

Every answer generated by KChat is sourced from a link in your knowledge base.

Your posts have two links, one to the KBucket popup and the other to the source URL.

Here, you can choose the outbound link for your chatbot.