GPT-6 Astra Computer Use in Action: From Blender 3D Models to Live Websites

AI is moving beyond answering questions and writing content. GPT-6 Astra computer use shows how AI can take part in practical tasks across software, files, websites and different applications.

These examples also show how AI workflow automation can connect several steps into one process. In this workflow, Astra is used to create a 3D model in Blender, share a recording with an editor and build a Next.js website from a reference image.

Watch the video below for the tutorial

What Is GPT-6 Astra Computer Use?

Computer use allows AI to interact with supported applications instead of simply telling you what to do.

With the right permissions, it can open applications, inspect what is on screen, click controls, enter information and move between tools while working towards a task. This makes AI desktop automation useful for work that involves more than one application.

Computer Use vs Browser Use

Browser use mainly happens inside websites and web applications.

Computer use can go further. For example, Astra may find a video stored on the computer, check whether it is the correct file and then open a browser to upload it.

This is where ChatGPT automation becomes more practical, because the workflow can move between local files, desktop software and online platforms.

Setting Up a Task in Codex

The workflow starts in Codex, where GPT-6 Astra is selected before giving the task.

Good instructions make a big difference. A Blender task needs a reference image. A file-sharing task needs the correct file and destination. A website task needs the design reference and preferred technology.

Clear inputs help AI automation tools understand the goal and reduce unnecessary corrections later.

Using Blender to Turn an Image into a 3D Model

One of the practical tasks uses Blender to create a 3D model from an uploaded image.

After access is approved, Astra works inside Blender using the image as a reference. It checks the model, makes changes and continues until a usable result is created.

This gives a practical example of how an AI 3D model generator workflow can work inside professional 3D software rather than simply creating another flat image.

Why the Prompt Matters in Blender

A reference image does not always show every detail needed for a 3D model.

That is why the instruction should explain what matters most. A simple concept model may only need the basic shape, while a detailed model may require more accurate proportions and smaller features.

Clear instructions help the AI produce a result that is closer to what is actually needed.

Finding the Correct Video

The next task is more familiar: finding the right recording from several files.

Instead of simply saying "upload the video," Astra is given clues such as the subject of the recording, its approximate time and where it is stored.

At first, the wrong recording is opened. Astra recognises the mismatch, closes it and then checks the correct Blender recording.

Uploading the Recording Through WeTransfer

Once the correct recording is found, Astra opens WeTransfer and starts preparing the file for sharing.

Some steps may still need human input, especially account details, verification or permissions.

This is a simple example of AI workflow automation where the AI handles repetitive steps while the user remains involved when approval is needed.

Sending the Link Through Microsoft Teams

The task does not stop after uploading the file.

Once the WeTransfer link is ready, Astra opens Microsoft Teams and sends it to the video editing group. Details such as the link expiry can also be included.

The complete ChatGPT automation workflow becomes:

Find the file → check it → upload it → get the link → send it to the correct group.

What Happens When the AI Makes a Mistake?

AI automation does not mean every action will be perfect the first time.

The wrong recording being opened is a useful example. Instead of continuing with the mistake, the workflow is corrected before the file is shared.

That is why important files, recipients and final outputs should still be checked.

Building a Next.js Website from an Image

Another task uses a website design image as the starting point.

Astra receives the reference and is asked to recreate the design as a Next.js page. The first version is reviewed, and further instructions are given to improve the layout, video content and animation effects.

This shows how an AI website builder workflow can move from a visual reference to an actual coded webpage.

From Website Design to Live Deployment

After the website is improved, the code is pushed to Git and deployed through Vercel.

The result can then be opened as a live website and checked for responsiveness, links, media and other interactive elements.

An AI website builder can speed up the development process, but the final page still needs proper testing before publication.

What These Tasks Show

These examples cover different types of AI desktop automation.

Blender shows AI working inside creative software. The recording task connects local files, WeTransfer and Microsoft Teams. The website task combines visual understanding, coding and deployment.

Together, they show how AI workflow automation can connect several actions instead of treating every step as a separate task.

Other Tasks AI Automation Tools Could Support

The same approach could be tested with spreadsheets, data entry and creative applications.

AI automation tools may also support workflows involving Premiere Pro, DaVinci Resolve and CapCut, depending on the available access and permissions.

The most suitable tasks are usually those with a clear goal and an outcome that can be checked easily.

Develop Practical AI Skills with GALTech School of Technology

Understanding AI becomes more useful when it is applied to real tasks.

GALTech School of Technology focuses on practical, industry-based training through guided projects, hands-on exercises and real-world AI applications. Students and professionals can build experience with modern AI tools while understanding how to plan, guide and review AI-assisted workflows.