Website Building SoftwareJust download the application and launch it.
Download here: https://rabitai.site/download/
All real-time chat messages are displayed here, including Twitch chat conversations, Twitch chat AI requests, user AI requests, automated AI interactions, and AI responses.
Chat messages are used to provide context when working with AI.
Ask the assistant
Is used when you want to send direct requests to the AI without using the microphone or Twitch chat.
The buttons below the chat are used to control the chat:
Clear all
Completely clears the chat.
Delete first
Deletes the first message in the chat.
Delete last
Deletes the last message in the chat.
AI queue
Count of AI requests currently in the queue for processing.
Clear queue
Clears the queue of AI requests awaiting processing.
This tab contains the main settings used to generate prompts when working with AI.
User Name
Used when sending a request to the AI in the Chat tab.
Bot Name
The name of your AI assistant (This only affects the text displayed, not the assistant's operation).
System Prompt
The main input field for the system prompt. This is where you enter the personality, appearance, algorithms, and other descriptions of your AI assistant.
Max Messages
The number of messages stored in the chat and used to generate context when working with the AI.
In the AI tab contains settings for connecting and interacting with the assistant.
The first block contains settings for connecting to Ollama:
Ollama URL
The Ollama API connection address. The default is a local address.
Model Name
The name of the model created in Ollama. The default is gemma3:4b.
The first request to the assistant will connect to Ollama using the specified credentials.
The second block contains settings for enabling and using AI memory:
Activate memory
Enables memory. When memory is enabled, chat is constantly saved to the AI memory.
Memory server address
The connection address to the mini-server that acts as the AI memory.
Always remember
Activates the mode in which the assistant accesses memory for each request to the AI.
Memory trigger
Trigger words. If these triggers are found in the AI request, the request to the AI memory is activated. Works when Always remember is disabled.
Connect
Connects to the memory server.
Disconnect
Disconnects from the memory server.
If Activate memory is enabled, the program will automatically connect to the memory server with the saved details when it is restarted.
The third block is settings for enabling and using internet search:
Activate internet search
Enables search queries.
Search triggers
Trigger words. If these triggers are found in the AI query, the internet search is activated.
This tab contains settings for interacting with the assistant via the microphone.
Speech recognition is implemented by transmitting the recorded audio track to an STT server compatible with the OpenAI API.
Activate STT
Enables STT.
STT Address
The address to connect to the STT server.
STT Model ID
The ID of the model loaded on the STT server.
STT API Key
The key used to secure connect to the STT server.
Activate hotkey
There are two options for interacting with the AI using voice: holding down a key and using a trigger word. When Activate hotkey is enabled, the first mode is enabled and the second is disabled.
Mic trigger word
When Activate hotkey is disabled, the program listens to the microphone in real time. If the specified trigger word is heard in your speech, voice recording will be activated.
Save trigger word
The trigger word will be included in the request when activated. This is useful if you address the AI by name.
Listen stop duration
The pause between words after which recording will stop.
Pre-recording
When enabled, app will take the specified number of seconds of pre-recorded audio to include in the request to the AI. This works for both key-press and trigger word modes.
Connect
Connect to the server.
Disconnect
Disconnect from the server.
If Activate STT is enabled, the program will automatically connect to the STT server using the saved credentials when it is restarted.
To facilitate interaction with the AI for multiple users, the STT server can now receive not only the recognized text but also the speaker's name, if the STT server allows it ("text": text, "speaker": speaker). If the server only returns text, the User Name from the General tab will be used.
This tab contains settings for activating your assistant's voice.
Voice-over is implemented by transmitting text to a TTS server compatible with the OpenAI API.
Activate TTS
Enables TTS.
TTS Address
The address to connect to the TTS server.
Model ID
The ID of the model loaded on the TTS server.
API Key
The key used to secure connect to the TTS server.
Connect
Connects to the server.
Disconnect
Disconnects from the server.
If Activate TTS is enabled, the program will automatically connect to the TTS server using the saved credentials when you restart it.
This tab contains settings for connecting and interacting with Twitch chat.
Status: disconnected
Clicking this will log RabiTAI into the Twitch website via a browser.
Enable chat command
Enables a Twitch chat command that will request Ollama when entered.
Chat command
The text of the Twitch chat command.
Request timer
A timer for the Twitch chat command. If you attempt to enter the command more frequently than the timer, a failure message will be sent to Twitch chat. The value can be set to 0, then the command can be entered without delay.
Timeout warning
The text of the message in Twitch chat about failure when entering a command more frequently than the value specified in the Request timer field. You can use the following variables in the text: {user} - the username from Twitch chat, {cmd} - the command from the Chat command field, {timer} - the value from the Request timer field, {timeleft} - the time remaining before the command can be entered again.
If the connection to Twitch is established successfully, when you restart the program, an attempt will be made to reconnect to the Twitch website.
The Vision tab contains settings for enabling and managing video capture from the camera.
Capture video
Enables video capture mode.
Trigger word
If enabled and a trigger is detected in the AI request, the current captured frame from the camera will be included in the request. If disabled, the captured frame will be included in every request.
Start capture at startup
When enabled, will start video capture when the program starts.
Capture device
Selects the camera for capture.
Video max width
Selects the video resolution in 16:9 format.
Capture every
The frequency with which a frame will be captured from the camera.
Start capture
Starts video capture.
Stop capture
Stops video capture.
If Capture video is selected, a frame will be captured regardless of whether video capture is enabled or disabled; in this case, a blank image will be sent. However, if Trigger word is enabled, a frame will only be sent when a trigger is detected in the AI request. If Start capture at startup is selected, video capture from the selected camera will be automatically activated when the program is launched.
This tab contains settings for connecting and interacting with the VTube Studio application.
Activate
Enables VTube Studio avatar mode.
VTube Studio API
The application connection address.
Connect
Connects to the application. The first time you connect, you'll need to confirm in the VTube Studio application. You have 60 seconds to confirm.
Disconnect
Disconnects from the application.
If Activate is enabled, the application will automatically connect to VTube Studio the next time you launch it.
The first section, Main Animations.
Action when listening
Allows you to select from a list a Hotkey action that will be triggered when a voice is recorded via STT.
Action when thinking
Allows you to select from a list a Hotkey action that will be triggered when the AI is processing requests.
Action when speaking
Allows you to select from a list a Hotkey action that will be triggered when the AI is speaking responses via TTS.
Action when idle
Allows you to select from a list a Hotkey action that will be triggered when the user is waiting.
You can select more than one Hotkey, in which case a random hotkey will be selected.
The second section, Triggerword animations.
Remove expressions on connect
Allows you to select a Hotkey action from the list. When enabled and selected, it is triggered every time you connect to VTube Studio.
+Add Trigger Word
Allows you to create a new Triggerword animation, for which you can select a Hotkey action from the list. The Hotkey action is triggered when a trigger is detected in the AI request.
Remove from TTS
When enabled, the trigger word will be removed from the text sent to the TTS server.
This tab contains settings for automatic assistant interaction with the user or Twitch viewers.
Enable interactions (Own Thoughts)
Enables automatic interactions. Own Thoughts is the user on whose behalf the request will be sent to the AI from the list below, while the AI will think that these are its own thoughts.
Minimum every
Minimum time for interactions.
Maximum every
Maximum time for interactions.
Interaction options
A list of requests to the AI that will be sent randomly at a random interval selected between Minimum every and Maximum every. Trigger words can be used in the text to launch internet searches, memories, and camera capture.
If Enable interactions is enabled, interactions will be automatically enabled when the program is restarted.
On the Simple bot tab, you can set commands for Twitch chat viewers, as well as automatic messages to send periodically to Twitch chat.
The first section, User commands.
Timeout warning
The text of the message to be sent to Twitch chat about failure if the command is entered more frequently than the value specified in the Timer field. You can use the following variables in the text: {user} - the username from Twitch chat, {cmd} - the command from the Chat command field, {time} - the time remaining before the command can be entered again.
Add command
Adds a new command to Twitch chat.
Text fields in order:
Command, Text to display for command, Timer.
The timer can be set to 0, in which case the command can be used without delay, or a value greater than zero, in which case the command can be used after a timeout.
Save
Saves the command. If you don't click it, the command won't be saved the next time you start the program.
Delete
Deletes the command.
When you restart the program, the saved commands will take effect automatically.
The second section, Timer commands.
Add timer
Adding a new timer message to the Twitch chat.
Text fields in order: Message text, Timer.
The timer cannot be less than 5.
Save
Saves the command. If you don't click it, the message won't be saved the next time you start the program.
Delete
Deletes the command.
When you restart the program, saved timer messages will be activated automatically.
This is the about page of the RabiTAI app. Here you can also find the current version of your app.
Website Building Software