Access the "Prompts (AI)" menu in the system to manage the prompts. Prompts are the artificial intelligences that users can interact with on the website. In the administrative system, you can add, modify, or delete a prompt.
On the homepage, you will see a list of prompts already pre-registered in the system. You can either use them as they are or simply delete all of them and start from scratch with your own prompts.
To register a prompt, click on the button on the right labeled "Add new AI."
Fill in all the required fields to add a new prompt, such as name, expert, slug, and training. Next, we will explore more about each of these fields.
Image of the Artificial Intelligence, we recommend sending it in JPG or WebP format, with a size of 640x700 pixels.
This field is used to specify the name of the AI.
Here, you can specify the expertise or domain in which the AI is knowledgeable or specialized.
The slug field allows you to define a URL-friendly identifier for the AI. It is typically used in the website's URL to access the specific AI.
It is important to highlight that the "slug" field must be unique. We usually avoid using accents and special characters in this field. Instead of spaces, you can fill it with "-".
In this field, you can provide a description or additional information about the AI. It helps users understand the AI's purpose or capabilities.
This information is also displayed on the chat page.
The welcome message field allows you to set the initial greeting or introduction message displayed when users interact with the AI. It is the first message users see from the AI.
This field allows you to assign categories to the AI training prompts. It helps organize and group the prompts based on specific themes or topics. You can add and manage categories in the "Prompts Categories" menu.
You can add one or more categories to the AIs (Artificial Intelligences). These categories are displayed on the home page, and you can enable or disable this feature in the settings. Additionally, the categories are also shown on the page that lists the AIs, acting as a filter to facilitate searching and organizing the available intelligences.
This parameter is responsible for defining the training of the intelligent AI agent. It is the text that the AI agent will use to introduce itself and identify as an expert in a certain subject.
For those more familiar with artificial intelligence, this field is the "prompt" or "input" of the AI.
For example, the cyber security expert Logan Frost has the following training:
"training": "Your name is Logan Frost. I want you to act as a cyber security specialist. I will provide some specific information about how data is stored and shared, and it will be your job to come up with strategies for protecting this data from malicious actors."
By writing in the training field, Logan will follow the provided instructions, including the responses he should provide about cyber security. Additionally, you can also specify negations, such as instructing Logan not to respond to questions outside the scope of cyber security. It is possible to define the tone that Logan will use when responding. For example, you can direct Logan to always respond in an objective or detailed manner.
By writing in the training field, you can define actions for the character and check their response. If you are not satisfied, you can modify the training field and continue testing until you get the desired result.
Improving a character's training depends on you: write in the training field, run tests, and check if you have met your expectations.
Temperature is a parameter used in language generation systems, such as those based on language-driven artificial intelligence (AI) models. This parameter is responsible for controlling the level of randomness in the responses generated by the AI.
A higher temperature will result in more creative and diverse responses, providing a wide range of possibilities in the answers. However, this higher level of randomness can also lead to less coherent or accurate responses, as the AI is exploring a broader variety of text generation paths. On the other hand, a lower temperature will produce more deterministic and predictable responses.
With a lower temperature, the AI tends to stick to safer paths, following more common and well-established lines of reasoning. However, this may result in less surprising responses since the AI will be less inclined to explore new paths or unconventional ideas.
The frequency penalty is a factor used during response generation by artificial intelligence to avoid word repetition in the generated answers.
This helps make the responses more varied and natural, encouraging the AI to use synonyms or different expressions instead of repeating the same word multiple times.
The idea behind this concept is to emulate how humans speak and write: we generally avoid repeating words in short spans of text, opting to use synonyms or restructure our sentences.
By applying a frequency penalty, the AI is encouraged to do the same, making its responses more diversified and, therefore, more similar to natural human text.
The presence penalty is another factor used during response generation by artificial intelligence. It penalizes words that are not present in the provided text during the response.
This means that the AI will be encouraged to use words and terms that have already appeared in the conversation, making the responses more relevant and coherent with the context.
This mechanism helps ensure that the AI stays "on track," avoiding significant deviations from the theme or context of the conversation.
Thus, the AI is encouraged to utilize the vocabulary and ideas that have already been introduced, increasing the probability that its responses are directly relevant and appropriate for the conversation's context. The presence penalty is, therefore, an important tool for maintaining coherence and relevance in the responses generated by the AI.
The system is divided into two audio parameters: system voices (free) and Google Cloud API voices (paid).
To enable the audio button in a chat, mark the option:
"Display voice icon in chat"
Text to Speech System Voices (Free)
Click on the "Check available system voices" option to display a list of available system voices. System voices are voices that are accessible within the browser.
Text to Speech Google Premium Voices (Paid)
You have the option to use Google's premium voices, offering more than 500 voices across various languages. Simply click the "Check available Google Cloud Voices" button to see the list.
Note: You will only be able to view the list and use Google's voices if you have a Google Cloud API key configured in the "Settings" -> "API key" field under Google Cloud Text to Speech API.
Learn more here:
https://cloud.google.com/speech-to-text
Learn more about pricing here:
https://cloud.google.com/text-to-speech/pricing
By enabling this function, you can embed the chat via iframe on another website. For example, if you have a blog or another site, you can place the chatbox for users to interact with the AI of your site within a post or page. We recommend using this option for non-VIP and free bots. To use this feature, simply register the bot first and select the "Allow embed chat (iframe)" option. Then, in the list of AIs, you will find a button to view the embedding code.
It is important to highlight that this function has some limitations:
DALL-E is an advanced artificial intelligence model developed by OpenAI, with the main purpose of generating images from descriptions provided by the user. For example, when enabling DALL-E for the AI, the user can make a request by including the "/img" command.
For instance: "/img white cat"
In response, DALL-E will create images of what was requested, specifically, images of white cats. This functionality allows the AI to produce visual representations based on specific commands, opening up diverse creative and practical possibilities for application in various fields.
You can enable or disable this function in the chat.
Chat min length is the minimum number of characters that must be typed in the chat box before the AI starts responding. This prevents overly short or uninformative responses, ensuring that the AI only begins generating answers when there is sufficient context.
Chat max length is the maximum number of characters allowed in the chat box before starting the conversation.
A boolean value (true or false) indicating whether the chatbot's avatar (image or graphical representation) should be displayed in the chat interface. The avatar can help give a visual personality to the AI and improve the user experience.
A boolean value indicating whether a button for copying chat messages should be displayed in the chat interface. This allows users to copy the content of the conversation, facilitating sharing or saving of interactions.
A boolean value indicating whether the AI should filter out inappropriate or offensive words from chat messages. This helps maintain a respectful and suitable conversation environment for users.
A boolean value indicating whether a list of contacts should be displayed in the chat interface. This list can contain information about other AIs or agents available for interaction, offering users the option to choose whom they want to chat with without needing to access the homepage or AI's page.
This is the API chat model used for this AI. Currently, you can use text-davinci-03, GPT-3, and GPT-4. We recommend using the gpt models, as the davinci model and other models will be discontinued by OpenAI.
Allows showing or hiding the chat sharing button for the AI. This button appears after the user's first interaction with the chat, enabling them to share the conversation through a publicly visible link.
This value represents the number of message history that the system will send to the OpenAI API to remember previous conversations. It's essential to consider that a larger history will consume more tokens and may potentially reach the API's token limit. We recommend carefully balancing this value.
Sometimes the message sent to the OpenAI API can be too long and needs to be truncated. This value represents the maximum character limit that the message should reach before being truncated.
Allows hiding or showing the AI output language selection box. If you decide to display the language selection box, the user can choose a language from the list. If you hide the selection box, you can still set a default output language for the text.
Allows hiding or showing the AI tone selection box. If you choose to display the tone selection box, the user can choose a tone from the list. If you hide the selection box, you can still set a default tone for the AI.
Allows hiding or showing the AI writing style selection box. If you decide to display the writing style selection box, the user can choose a writing style from the list. If you hide the selection box, you can still set a default writing style for the AI.
Allows you to show or hide the AI suggestions button. If activated, this option can help users by providing suggested questions they can ask the AI.
Tip: If you have any doubts about how to fill in a specific field, you can simply edit an existing AI and check how all the fields have been filled. This will provide you with a clear reference on how to properly fill in each field.
