Set up a local AI assistant with NVIDIA ChatRTX on your PC

Download the ChatRTX installer

Open your preferred web browser, like Chrome, Edge, or Firefox. In the address bar, type "NVIDIA ChatRTX" and press Enter, then look for the official NVIDIA product page. Once you're on the ChatRTX product page, you'll see information about the tool and usually a prominent download section. Look for a button clearly labelled something like Download for Windows or Get ChatRTX. Click this button to begin downloading the installer file.

What you see: Your browser will likely show a progress bar or a notification indicating a file is downloading, often at the bottom of the browser window or in a dedicated downloads panel. The file will typically be named ChatRTX_Setup.exe or similar. What happens next: The installer file, which is essentially a small package containing all the necessary setup instructions for the program, will begin downloading to your computer's Downloads folder. This might take a few minutes depending on your internet connection. If it looks different: If you click the button and nothing seems to happen, check your browser's download history or your Downloads folder to ensure the download didn't start automatically or get blocked. Sometimes, a popup might ask for permission to download; click 'Allow'. You'll know it worked when: The download completes, and you see the ChatRTX_Setup.exe file (or a similar name) clearly visible in your browser's download list or directly in your computer's Downloads folder.

Run the installer and agree to the terms

Navigate to your Downloads folder, where you saved the installer file from Step 1. Find the ChatRTX_Setup.exe file and double-click it to start the installation process. A small window will appear, asking you to grant permission for the installer to make changes to your device; click Yes. Next, you'll see a screen displaying the NVIDIA Software License Agreement. Read through it if you wish, then locate and tick the box that says I accept the terms (or similar wording). Finally, click the Install button to proceed.

What you see: After granting permission, a new window will open showing the license agreement. Following your acceptance, you'll see a progress bar indicating the installation is underway, copying files and setting up the program. This process can take several minutes as it prepares everything for ChatRTX. What happens next: The installer will copy all the necessary files to your computer and configure the software. It will show a progress bar or a series of messages. Once complete, you should see a message confirming Installation Complete or similar. If it looks different: If the installer asks for administrator permissions multiple times, always click Yes to allow it to continue. If the installation seems to freeze, try waiting a few more minutes, as it might be downloading additional components in the background. You'll know it worked when: The installer window closes automatically, and a new shortcut icon, usually labelled NVIDIA ChatRTX, appears on your desktop, ready to launch.

Launch ChatRTX and sign in

Locate the new NVIDIA ChatRTX shortcut icon on your desktop. Double-click this icon to open the application for the first time. The ChatRTX window will appear, and it will typically prompt you to sign in to your NVIDIA account. You'll likely see a clear button or link that says Sign In or Log In. Click this button and then enter the email address and password associated with your free NVIDIA account in the fields provided.

What you see: The ChatRTX application window will open, often with a large welcome screen. You'll see fields for your email and password, and a 'Sign In' button. Once you've entered your details and clicked 'Sign In', the screen might briefly show a loading spinner. What happens next: After successfully signing in, the application will fully open to its main chat interface. You'll be greeted with a blank chat area or a brief welcome message from the AI, and crucially, a blinking cursor at the bottom of the screen, indicating it's ready for you to type your first message. If it looks different: If you don't immediately see a "Sign In" button, look for a small person-shaped icon, a gear icon, or a menu item (often in the top-right corner) that might lead to account settings or a login prompt. Ensure your computer is connected to the internet during this step, as it needs to verify your account. You'll know it worked when: You see the main chat window, with a text box at the bottom where you can type messages, and no further prompts asking you to sign in.

Pick your assistant model

Once ChatRTX is open, you need to choose the "brain" (called a model) that your AI assistant will use. A model is essentially a pre-trained set of instructions that allows the AI to understand your questions and generate responses. Look towards the top-left corner of the ChatRTX window. You'll find a dropdown menu, likely labelled Select Model or similar. Click on this dropdown menu to see a list of available AI models. For a good balance of performance and capability, we recommend choosing mistral-7b-instruct-v0.2.Q4_K_M or llama-3-8b-instruct.Q4_K_M if it's available. The 'Q4_K_M' part just indicates the file size and efficiency of the model.

What you see: A dropdown list will appear, showing various AI model names. Selecting one will typically cause the chat window to dim slightly or show a progress bar. What happens next: Once you've clicked on your chosen model, the application will begin to load it into your computer's GPU memory. This process can take a minute or two, especially the first time, as the AI's "brain" is being activated. You might see a message in the chat area like Model loaded once it's ready. If it looks different: If the dropdown menu is greyed out or unresponsive, try closing and re-opening the ChatRTX application. Ensure your GPU drivers are up to date, as this can sometimes affect model loading. If a specific model won't load, try picking a smaller one from the list to test functionality. You'll know it worked when: The chosen model's name appears next to the 'Select Model' dropdown, the chat prompt at the bottom of the screen is active, and you see a confirmation message like "Model loaded" in the chat history.

Teach it about your home

Now that your AI assistant is ready, let's give it some context about your life so it can be more helpful. Since ChatRTX runs locally, all the information you share stays on your computer, making it very private. In the chat box at the bottom of the ChatRTX window, where the blinking cursor is, type a few sentences describing your home environment and daily routine. The more details you provide, the better it can assist you.

What you see: The chat box will be empty, waiting for your input. As you type, your message will appear in the box. What happens next: After you press Enter, ChatRTX will process your message. It will then reply, usually by summarising or acknowledging the information you've provided, confirming it has "learned" these new details. This information is stored locally and should be remembered for future chats. If it looks different: If the AI's response seems to miss the point or doesn't acknowledge the facts you've given, try rephrasing your input more clearly, perhaps using bullet points or simpler sentences. You'll know it worked when: The assistant replies with a message that confirms it has understood and will remember details about your home, such as "Okay, I've noted you live in a two-bedroom apartment in Sydney, with specific smart devices and a routine."

Ask your first question

With your AI assistant having some knowledge of your home, you can now start asking it practical questions. Go back to the chat box at the bottom of the ChatRTX window. Think of something you might usually look up online or ask a housemate. Type your question clearly and press Enter.

What you see: Your previous conversation with the AI, including the information you gave it about your home, will be visible. Your new question will appear in the chat history after you type it. What happens next: The AI assistant will process your question, using the contextual details you provided in the previous step (like your location in Sydney) to give you a more personalised and practical answer. It will type out its response directly into the chat window. If it looks different: If the answer seems irrelevant or incorrect, don't worry. This is common with AI. Try rephrasing your question to be more specific, or simplify it if it was too complex. For example, instead of "What time should I leave for Bondi?", try "What time should I leave my apartment to get to Bondi Junction by 9am if I walk to the station?" You'll know it worked when: You receive a sensible, relevant, and practical answer from the AI, which might even incorporate the details you shared about your location or routine.

Connect a smart device (optional)

One of the powerful features of ChatRTX is its ability to interact with certain smart home devices, allowing you to control them using natural language. If you have a compatible smart light, thermostat, or similar device, you can attempt to connect it. Look at the very top of the ChatRTX application window for a menu bar. Click on the Tools option in this menu, and then select Device Manager from the dropdown. Inside the Device Manager, look for a button or option like Add Device or Connect New Device. Follow the on-screen prompts to link your smart home device to ChatRTX.

What you see: Clicking 'Tools' will reveal a dropdown menu. Selecting 'Device Manager' will open a new panel or window, often showing a list of connected devices (likely empty at first) and options to add new ones. What happens next: If the connection is successful, your device will appear in the Device Manager list. You can then close the Device Manager and return to the chat window to give commands. For example, you could type "Turn on the living room light." If it looks different: If you don't see a 'Tools' menu, or if 'Device Manager' isn't an option, it's possible this feature isn't supported on your version of ChatRTX or your specific GPU. Not all smart devices are compatible; if yours isn't connecting, this feature might not be available for your setup. You'll know it worked when: Your smart light or other device responds to your command in the chat, actually turning on or off as requested, and you see a confirmation message from the AI.

⚠️ Device support depends heavily on your specific hardware and smart home brand. Not all devices are compatible with ChatRTX, and some integrations might be experimental.