How to use ChatGPT as a native assistant in Windows 11 and integrate AI into your desktop

If you use Windows 11 and you'd like to have an AI assistant always at hand on your desktop, you may have noticed that Microsoft is heavily promoting its own Copilot.However, not everyone is thrilled with the system's latest AI features, such as the controversial Recall, or with the idea of ​​being tied to a single assistant.

The good news is that it's possible today. integrate ChatGPT as a near-native assistant on your PCYou can call it with keyboard shortcuts, use it from an official desktop app, or even opt for entirely local solutions that don't send your data to the cloud. We'll explore all the options: from configuring the Copilot key to launch ChatGPT to setting up a private AI with GPT4All or using "Jarvis"-type tools that run code on your machine.

Copilot, Recall and the controversy surrounding AI integrated into Windows 11

The latest major Windows 11 updates have arrived packed with AI-based features that haven't pleased everyoneThe most talked-about case has been Recall, a feature capable of recording almost everything you do on the device to allow you to later search for "anything you saw on the screen." The idea sounds powerful, but many users have found it to be a complete invasion of their privacy.

In addition, Microsoft has been working on this for some time. Putting AI at the heart of the Windows experienceCopilot has become the company's big bet, with an increasingly visible presence in the taskbar and, above all, with the arrival of physical keyboards that include a dedicated Copilot key, at the same level as the Windows key.

The problem comes when You don't want to use Copilot, but you do want AIUntil recently, if you pressed the Copilot key or the Win + C shortcut, there was no other option: Copilot would appear no matter what. There was no official way to change this behavior to launch another assistant application. This lack of freedom has been one of the most frequent criticisms from the community.

Given this context, Microsoft has begun to loosen restrictions slightly. New Windows 11 updates now allow this. customize which AI assistant is launched with the Copilot key and with Win + CAnd this is where ChatGPT comes in, which can become your preferred assistant if you so choose.

Configure the Copilot key and Win + C to use ChatGPT in Windows 11

The most interesting new feature for those who want to integrate ChatGPT into their desktop is that Windows 11 now allows you to replace Copilot as the default assistant by pressing the Copilot key or the Win + C shortcut. For now, official compatibility focuses on ChatGPT, but the foundation is laid for other assistants to be added in the future.

The change came with Recent updates to Windows 11 24H2 (KB5058499) and 23H2 (KB5058405)If your system is up to date, you should see a specific option within Settings to tell Windows which assistant to launch when you request "smart help" from the keyboard.

What this function does is redirect the Copilot key or Win + C event towards an AI assistant application that meets Microsoft's requirements. In other words, you can't use that shortcut for just any program, only for apps recognized as AI assistants. This is where the official ChatGPT app for Windows comes in.

Once configured, every time you press the Copilot key or Win + C, instead of seeing the Copilot panel embedded in the system, it will open the ChatGPT windowIt's a very convenient way to have the chatbot ready to respond at any time, without going to the Start menu or opening the browser.

Install the ChatGPT app for Windows from the Microsoft Store

Before you can choose ChatGPT as the default assistant in Windows settings, you need Install the official ChatGPT app for WindowsThis desktop application is now publicly available to Windows 10 and Windows 11 users, and is not limited to those who pay a subscription fee.

Installation is simple: simply Open the Microsoft Store and search for “ChatGPT” Or you can go directly to the app's official link if you have it handy. From the product page, simply click "Install" and let Windows complete the process as you would with any other application.

Once installed, you'll have a ChatGPT icon in your Start menu and you'll be able to Open the program like a normal desktop appFrom there, Windows will be able to detect that there is an available AI assistant that meets the requirements to replace Copilot by pressing the corresponding key or shortcut.

In addition, this ChatGPT app for Windows incorporates its own global shortcut, Alt + SpaceWhen you use it, a floating window opens from which you can talk to the model, ask chained questions, start new conversations, upload screenshots, or attach images and videos for the AI ​​to analyze.

For users with paid plans, the application also provides access to advanced features such as Advanced Voicewhich allows you to chat with ChatGPT by voice in real time, directly from your computer. All this without depending on the browser, making the experience much more integrated into the desktop.

How to change the default assistant from the Copilot key to ChatGPT

With the ChatGPT app already installed, the next step is to go to the Windows 11 configuration to tell the system to stop opening Copilot When you use the dedicated key or Win + C. The setting is somewhat hidden, but it can be changed in a few seconds.

The first thing to do is open the Settings app and go to the section "Personalization"Within that menu you will find several options related to the appearance and behavior of the system, and one of them is focused on how the keyboard and text input in general work.

Now enter the section called “Text entry”Inside you will see, among other things, a specific option with a name very similar to "Customize the Copilot key on the keyboard" or equivalent, depending on the exact version of Windows 11 you have installed.

By expanding that option, the system offers you a behavior selectorBy default, it's usually configured to open Copilot. What you need to do is choose the option marked "Custom" or similar, which allows you to assign the key to a different assistant.

If the ChatGPT app is open or installed correctly, a pop-up window for you to choose which assistant you want to useChatGPT should appear in that list as a compatible option. Simply select it, confirm, and let Windows save the change. From then on, pressing the Copilot key or Win + C will launch ChatGPT instead of the Microsoft assistant.

What can you do with ChatGPT as an assistant on the Windows desktop

Once you have ChatGPT assigned as your default assistant, the interesting thing is that You can summon him no matter what you're doing. on your PC. Whether you're in the browser, in Word, playing games or programming, a tap of the Copilot key or the Win + C shortcut (or Alt + Space, depending on whether you're using the Windows integration or the app's own integration) will open the chat.

With the desktop application you can, in addition to text messaging, upload screenshots for the AI ​​to interpretDrag photos and videos to request descriptions or analyses, or perform quick queries without having a hundred tabs open in your browser. It also integrates with internet access to respond with updated information when the model allows.

For many users, this makes ChatGPT a versatile assistant for daily tasks: explaining errors in system messages, summarizing documents you have open, generating emails, helping with Excel formulas, preparing texts, or even giving ideas while you write or study.

If you opt for payment plans, things go a little further thanks to The most advanced models already have voice functionsBeing able to talk to the computer in real time, without having to type, brings the experience even closer to that "Jarvis" that many imagine, especially if you combine the desktop app with other tools.

It must be borne in mind, however, that today Microsoft limits the use of the Copilot key and the Win + C shortcut exclusively to AI assistantsYou cannot reuse that shortcut to open any generic program. The intention is for that shortcut to remain a direct gateway to artificial intelligence solutions within their ecosystem.

Differences between Copilot and ChatGPT as assistants in Windows 11

Although Copilot and ChatGPT share a technological base thanks to the close collaboration between Microsoft and OpenAI, when it comes down to it the user experience is differentChoosing one or the other has more to do with your preferences, your workflows, and the extra features you need.

Copilot is more integrated into the Microsoft ecosystemIt has direct access to services like Edge, Bing, and, in professional environments, Microsoft 365, allowing for more in-depth work with Word documents, Excel spreadsheets, or PowerPoint presentations, depending on the plan subscribed to. It's the system's "official" assistant, designed to enhance Microsoft's offering.

ChatGPT, on the other hand, offers a a more neutral experience focused on the model itselfFrom its Windows app, you can take advantage of the chatbot's continuous improvements, its multimodal capabilities, and, in paid plans, advanced options that don't always fit within the Copilot framework. For many users, it's also more convenient and straightforward to work "with ChatGPT" than with an additional branding layer.

In terms of response quality, the real difference will depend on which specific model are you using in each case (for example, GPT-4 variants or lighter models). As a general reference, the experiences are similar in terms of power, because both draw from the same line of research and development, but the way they integrate with the system and surrounding applications is what determines the final experience.

Ultimately, being able to choose which assistant starts with the Copilot key and the Win + C shortcut means that You are no longer locked into the default optionThis opens up an interesting possibility for other players, such as Perplexity or even Google's Gemini, to offer desktop clients that integrate in the same way and compete head-to-head to be the brain of your PC.

Use ChatGPT as “Jarvis” to run code and tasks on your PC

Beyond the typical chat conversation, one of the ideas that most appeals to enthusiasts is turn ChatGPT into a kind of domestic “Jarvis” capable of performing real actions on the computer: creating and modifying files, launching scripts, automating small tasks, or helping with system administration.

For this type of use, many users resort to third-party tools such as Open Interpreter or similar projects. The idea is connect the OpenAI API to an environment that can execute code in a controlled manner (for example, on Linux Mint, Windows or macOS), so that the AI ​​writes scripts and the system itself executes them under the user's supervision.

The typical flow is that the AI ​​generates code in Python, Bash, or other languages, and the interpreter Run it locally with limited permissionsThis allows AI to read and modify files, obtain hardware information, analyze logs, or automate repetitive tasks. It's a very powerful approach, but one that must be handled carefully, because any error in the script can affect the system.

One of the common problems when using this approach with the OpenAI API is encountering the limits of the models' useFor example, powerful models like GPT-4o can quickly exhaust the request or token limit if programming is intensive, causing tools like Open Interpreter to crash with error traces when the maximum allowed speed is reached.

In these cases, some users choose to switch to lighter models, such as the GPT-4 mini or other more economical variantsTo reduce resource consumption and stay within free or low-cost limits, even these models can encounter speed limits under heavy use, and not all tools elegantly handle automatic retries or dynamic model switching.

Choosing models and alternatives when there are API speed limits

If you're going to build a "Jarvis"-type assistant that writes a lot of code, it's advisable Choose the right model based on your needs and the available budget. Without a comprehensive payment plan, speed limits and token consumption are key factors that restrict what you can do.

A common strategy is to reserve the most advanced models for critical tasks. complex reasoning or significant code refactoringFor more repetitive or simple tasks, lighter, less energy-intensive versions are used. This allows for extended usage without completely sacrificing quality when it's truly needed.

When a tool like Open Interpreter encounters a speed limit, ideally it should Implement retries with progressive waiting or that automatically switches models. If you're setting up the environment yourself, it's worth reviewing the application options to adjust the default model, request cadence, and boundary error handling.

In the absence of a perfect solution, many users combine the use of ChatGPT via API with other suppliers or with local models that run on their own machine. Thus, if one service is temporarily blocked by limitations, they can continue working with another AI engine, even if it is less powerful or does not have all the multimodal capabilities.

It is also important to control the amount of context sent in each API call, because sending unnecessarily long chat histories This increases the total number of tokens consumed and accelerates reaching the limits. Designing more concise prompts and breaking down complex tasks into smaller steps helps reduce this problem.

Install and use GPT4All: a local and private AI for your PC

If your priority is privacy and you don't want your data is constantly being sent to external serversAlternatively, you can opt for a local solution like GPT4All. This open-source project allows you to run language models directly on your computer, without needing an internet connection once the models are downloaded.

After completing the installation, the first time you open the application, the following will be displayed: a window to choose the models you want to downloadEach template takes up several gigabytes, as it comes pre-programmed and packaged to work independently of the cloud. You can select one or more templates and click "Download" to begin the download.

Once the templates have finished downloading, simply Close or minimize that window and access the main GPT4All interface. At the top you will see a drop-down menu from which to select which model to use at any given time, from all those you have installed locally on your computer.

From there you can start writing questions and receiving answers, just like in a regular chat. These models have been trained to Answering general questions, helping with writing, summarizing documents, and performing simple tasksHowever, their capacity is generally lower than cloud solutions like GPT-4. In return, they gain in privacy and control over data.

Advantages and disadvantages of using local models like GPT4All

The great advantage of GPT4All is that It does not need a permanent internet connection to function Once the templates are downloaded, the conversations and data you work with remain on your machine, which is very useful if you handle sensitive information or simply prefer to minimize sending data to external servers.

The downside is that most of the models available on GPT4All are based on previous architectures such as GPT-J (derived from GPT-2) or in variants of LLaMAAlthough they have been improved and refined, their level of text comprehension and generation is usually lower than that of more modern commercial models that run in the cloud.

This is especially noticeable in very complex tasks, lengthy reasoning, or advanced programming problems. For general inquiries, help with texts, relatively simple summaries, and day-to-day tasks, They are usually more than enoughprovided you adjust your expectations and accept that you won't get the same performance as with the latest versions of GPT-4.

Another point to consider is that many of these models are primarily trained in EnglishAlthough they understand and produce Spanish, they can sometimes mix languages ​​or give somewhat noisy responses when pushed too hard in Spanish. Even so, for everyday use and as support for personal projects, they work reasonably well.

The GPT4All application also includes a small menu from which update the app itself or download new models that the project developers will publish. It's a constantly evolving tool, and it's expected that in the future it will even allow you to train your own models with custom datasets.

This whole picture shows that right now you have a very wide range of ways to integrate AI into your Windows 11 desktopFrom using ChatGPT as a near-native assistant, replacing Copilot with a dedicated key and Win + C, to the official desktop app with global shortcuts and voice functions, to building your own "Jarvis" that executes code via the API, or opting for entirely local solutions like GPT4All to prioritize control over your data. Depending on how much you value Windows integration, the power of the model, or control over your data, you can combine these options to make artificial intelligence work in your favor without having to settle for the default features.