NVIDIA unveils how G-Assist is a game-changer for PC gaming and content creators

NVIDIA showcased some impressive AI technologies at Computex 2025, with the company leading the charge of AI integration into the gaming space. One of those technologies is Project G-Assist, and I got a behind-the-scenes demo of what is new with the feature. For those who don't know, it is a local AI agent designed to enable a user to control their PC through voice/text chat.

G-Assist has received some considerable improvements, with NVIDIA demonstrating the AI agent's ability to launch titles through a user activating the microphone and simply saying, "Open Counter-Strike 2." Moreover, G-Assist, currently available to download within the NVIDIA App, can also optimize a game's settings without user intervention. 

A user will interact with G-Assist either through text chat or via a microphone and simply ask it to optimize the settings for a game. The AI agent will go ahead and optimize the in-game settings based on the PC configuration, improving average framerate.

2

Once a game has been launched, G-Assist is able to take advantage of the NVIDIA Overlay features, such as recording gameplay, showing performance metrics, capturing replays, and even charting CPU and GPU performance. NVIDIA demonstrates how a user can just say, "Tell me what my GPU and CPU utilization are."

G-Assist responded extremely quickly with the respective utilization numbers, which NVIDIA followed up with, "Show me a graph of my CPU and GPU utilization over the last 30 seconds". G-Assist responded with a graph showcasing both metrics over the requested time period. Impressive stuff for those interested in benchmarking their rig.

Practical use in Games

NVIDIA showcased how G-Assist can interact with different chats and services gamers typically use. Since G-Assist now supports API plugins, NVIDIA demonstrated how a user can take advantage of G-Assist to interact with software such as Discord without having to manually click on the Discord window. Moreover, a user can call upon the Discord plugin with various requests.

NVIDIA demonstrated a user informing their Discord community they would be online in five minutes with the following request: "Hey Discord, tell my chat I'll be online in five minutes." G-Assist proceeded to send that message to the Discord community. G-Assist was also used to send a gameplay recording to Discord.

The possibilities don't stop there, as G-Assist has a plugin for Google Gemini, meaning users can request information from Google's AI chatbot, enabling information to be pulled from Google Search. NVIDIA showed how a user could call upon information about the game they are playing, while it's being played.

NVIDIA showed how G-Assist used Gemini to display the scheduling for the next Counter-Strike 2 competitive matches - which it easily did. Imagine using G-Assist for information about a boss you are currently struggling with, instead of tabbing out to a search engine and hunting for information on the boss that way.

NVIDIA also demonstrated how G-Assist was able to keep in context of the ongoing conversation about Counter-Strike 2, with the AI agent being asked, "Who is the favorite to win the competition?" After pinging Google's Gemini API, G-Assist responded with a long answer, but eventually put its money on Team Vitality. 

Since Spotify has an API, G-Assist can start playback on Spotify and also control a system's RGB lighting through SignalRGB. But what if the software you want to interact with doesn't have a plugin available to download? Well, NVIDIA has thought of that, and the main problem is that not everyone knows how to code a custom plugin. The solution is a ChatGPT-dedicated G-Assist Plugin builder that can be interacted with through natural language to create custom plugins that will work with G-Assist.

All that is required is that the software that the user is trying to interact with has its own API available to allow G-Assist to send commands to it. Moreover, users don't even have to create their own plugins if they don't want to, as custom plugins can be uploaded to the community page, where users can browse and download plugins developed by the community.

Using G-Assist to Share Gameplay to Social Media

I asked NVIDIA if it was possible to enable G-Assist to post replays/recorded gameplay to social media platforms such as X, formerly Twitter, since it was able to share a recording directly with Discord. NVIDIA responded by saying absolutely. All that is required is that the target social media platform has an API that enables posting. If it does, a plugin for G-Assist can be created, and the process utilized.

I can imagine G-Assist being used by a streamer who wants to instantly publish a stream highlight, possibly captured with NVIDIA's own Shadowplay or Replay feature, to their social media platforms. The current struggle is that if a streamer wants this published immediately, they have to stop the stream momentum by pausing the game, finding the clip, and uploading it to each social media platform separately.

With G-Assist, a user could simply say the following, "Hey X, upload the most recent highlight with this caption," and the local AI would carry out the process for the content creator without the game ever being paused, which maintains a better experience for both the creator and viewers as the streamer can continue on with being entertaining, while simultaneously posting a stream highlight that could in-turn attract even more viewers - if the creator included a link to their stream within that G-Assist post.

The most powerful aspect of G-Assist is its ability to communicate with any API. Users can communicate with several APIs simultaneously through the use of custom ChatGPT-powered plugins, with NVIDIA saying to me that it has essentially created an API calling AI agent. The possibilities here are truly endless.