Editor, encoder and custom GPT

What if collaborating with ai could feel as natural as working with a teammate? That's what OpenAI's new feature, Canvas, is for! Released on the fourth day of the 12 Days of OpenAI series, Canvas is now available to all ChatGPT users. Whether you're creating a compelling story, debugging complex code, or exploring new ideas, Canvas offers a parallel workspace that perfectly fits your needs. Let's dive into what makes this tool special.

<blockquote class="twitter-tweet”>

Canvas, a new way to work with ChatGPT to compose, edit, and get feedback on writing and code, is now available to all users of our 4o model.

It is fully implemented on the web and in the ChatGPT desktop app for Windows. pic.twitter.com/1MVvmXphZM

—OpenAI (@OpenAI) <a target="_blank" href="https://twitter.com/OpenAI/status/1866578914233159928?ref_src=twsrc%5Etfw”>December 10, 2024

How to access Canvas?

1. Log in to your ChatGPT account via the web interface.
2. In the chat composer, click the tools icon to see all available tools.
3. Choose Canvas from the list. This will open a new workspace where you can start creating content or coding.

Canvas Key Features

1. Wide accessibility– Canvas is now integrated directly into the main ChatGPT interface and is accessible to all users, regardless of their subscription plan.
2. Python code execution: Users can now run Python code directly within Canvas, complete with text or graphical output, ensuring a seamless coding experience.
3. Integration with custom GPTs– Canvas now fully supports custom GPTs, allowing developers to use the potential of this tool in their custom ai models.

I will explore these features in the next section.

Let's use ChatGPT Canvas

To help you get started, here are some practical examples of how you can use Canvas:

Wide accessibility

Canvas introduces a split view format that allows users to work simultaneously with the chat and the document. This eliminates the clutter of edits and revisions in the chat feed, allowing for clearer, more structured interaction.

Example

I evaluated the content of my article with ChatGPT and used the wide accessibility option. After pasting my content into the message window, I zoomed in:

This is what the enlarged screen looks like:

editing tool

You can edit documents directly within Canvas, apply formatting, and even receive suggestions from ChatGPT in real time. It's pretty simple to use: copy the content of the article, paste it into the message box, open it in Canvas mode, and select editing mode. Add prompts based on the feedback you need and Canvas will suggest modifications.

Example

I used a blog about “Generative ai Skills for Engineering Students” and gave Canvas the following message:

Think like an editor and add suggestions on how to divide this content into different titles. While posting this article on my blog, please suggest what keywords can be added to improve its searchability.

Production:

ChatGPT provided valuable suggestions right in the Canvas window. When I clicked the Apply button, the changes were incorporated without any problems. This feature makes editing and maintaining formatting much easier, even when transferring content to Google Docs.

Improved feedback mechanism

A notable feature is ChatGPT's ability to provide contextual feedback on user-supplied text. This is particularly valuable for writers looking for detailed feedback or students who want to polish academic essays.

Example

For the same article, I wanted to check if it covered everything a student needs to learn about generative ai. I asked Canvas:

Think like a student planning a career in GenAI. Is this item really beneficial? Add comments accordingly.

Canvas provided valuable suggestions that could help students improve their learning experience.

Code with canvas

For programmers, Canvas offers real-time debugging and code execution within the editor. This includes syntax highlighting, error detection, and visual results for data science tasks.

Example

I tested a buggy code snippet and handed it to Canvas with this message:

Why is my code not working?

import matplotlib.pyplot as plt

import numpy as np

# Create data

x = np.linspace(0, 10, 100)

y = np.sin(x)

# Create a figure with publication-quality aesthetics

plt.figure(figsize=(6, 4), dpi=300)  # Set figure size and resolution

plt.plot(x, y, label="Sine Wave", color="b", linewidth=2)

# Adding labels and title

plt.xlabel("x-axis label", fontsize=14)

plt.ylabel("Y-axis label", fontsize=14)

plt.lable("Sine Wave Example", fontsize=16)

# Adding grid and legend

plt.grid(True, which="both", linestyle="--", linewidth=0.5)

plt.legend(fontsize=12)

# Saving the figure as a high-resolution PNG

plt.savefig("sine_wave_figure.png", dpi=300, bbox_inches="tight")

plt.show()

Canvas identified the errors and allowed me to fix and run the code in the right-hand window. This feature is invaluable for programmers!

The Fix Errors option in the Canvas window rewrites all code line by line to ensure there are no errors.

Edited code on canvas:

import matplotlib.pyplot as plt
import numpy as np

# Create data
x = np.linspace(0, 10, 100)
y = np.sin(x)

# Create a figure with publication-quality aesthetics
plt.figure(figsize=(6, 4), dpi=300)  # Set figure size and resolution

plt.plot(x, y, label="Sine Wave", color="b", linewidth=2)

# Adding labels and title
plt.xlabel("x-axis label", fontsize=14)
plt.ylabel("Y-axis label", fontsize=14)
plt.title("Sine Wave Example", fontsize=16)

# Adding grid and legend
plt.grid(True, which="both", linestyle="--", linewidth=0.5)
plt.legend(fontsize=12)

# Saving the figure as a high-resolution PNG
plt.savefig("sine_wave_figure.png", dpi=300, bbox_inches="tight")

plt.show()

Production:

Custom GPTs with Canvas

Integrating Canvas with custom GPTs opens up a whole new range of possibilities. As part of this initiative, Canvas will become a default tool for new custom GPTs, while existing GPTs will be able to enable it through their configuration settings.

Previous OpenAI Releases

What's next?

Canvas is a fun and interactive way to work with ai. It makes working on projects feel natural and productive, whether you're a writer polishing an article, a coder debugging scripts, or an educator designing lessons. OpenAI has made it simple and accessible, encouraging users to dive in and see what they can achieve.

As OpenAI continues to improve Canvas, it's clear that this is just the beginning. The tool promises to become an even more essential part of working with ai, making our interactions more fluid and effective.

So why wait?

Explore Canvas, share your experiences with me in the comments section below.