Generative AI

ChatGPT branches out: the “Create a Branch” tool paves the way for smarter navigation

In everyday use, generative artificial intelligence relies largely on a seemingly simple interface: a dialog box. This linear format, inherited from chat applications, nevertheless limits certain complex interactions, particularly when it comes to testing different hypotheses, revisiting a previous point, or exploring multiple options simultaneously.

With the quiet arrival of the “Create Branch” button on ChatGPT, OpenAI is introducing a major conceptual innovation1 : allowing users to branch off a conversation from a given point without overwriting the main history. Behind this feature lies a paradigm shift: moving from a linear conversation model to a tree-like navigation structure, better suited for professional, creative, or educational uses.

The "Create a branch" button now appears in certain versions of ChatGPT (specifically for users of the Plus version or those testing via GPT-42). It appears next to the messages generated by the AI. By clicking on it, the user opens a new conversational thread, separate from the main thread but starting from the same point.

In practical terms, this means that you can:

  • to develop a different hypothesis based on an intermediate question or prompt,
  • test several formulations or approaches without losing track of the rest of the discussion,
  • create variations of a text, code, or plan, while maintaining a structured record.

Each thread can then develop independently, while remaining linked to the original conversation.

While this feature may seem minor at first glance, it marks a significant step forward in the design of AI interfaces. Until now, users had to either:

  • edit an existing message, which replaced the rest,
  • start a new discussion, at the risk of losing the context,
  • or browse through the history, without being able to create variations within the same thread.

With “Create a Branch,” the interaction takes on a tree-like structure. This approach aligns with the principles used in software development (Git), mind mapping, and narrative design (interactive scenarios). The user becomes a co-architect of their journey with the AI.

This new structure offers exciting opportunities for a variety of users:

  • Exploratory research: the ability to explore different avenues based on a single initial question; useful in strategy, market intelligence, or report writing.
  • Education and training: A teacher can generate multiple versions of a correction, or a student can compare different approaches.
  • Programming: testing different code architectures, or comparing functional and object-oriented approaches, without having to start from scratch.
  • Creative Writing: Writing interactive stories or comparing literary styles based on the same opening.

By allowing everyone to organize their interactions through trial and error, the “Create a Branch” tool fosters a form of AI-assisted divergent thinking.

This feature is part of a broader evolution of ChatGPT, which has already begun with:

  • custom instructions, which allow you to set up a user profile,
  • specialized tools (code, data, browser), accessible depending on the context3.
  • and browsing the history, which keeps conversations in separate "threads."

With “Create a Branch,” OpenAI is adding a key building block that transforms ChatGPT from a simple conversational assistant into a true, modular, and structured environment for intellectual exploration.

Despite its strengths, the feature still has room for improvement. Three limitations stand out:

  1. Readability: As the number of branches increases, the interface can become difficult to follow without a clear graphical representation.
  2. Accessibility: This "branch" approach isn't always intuitive for users who aren't familiar with technical tools.
  3. Backup and navigation: It is not yet possible to view all branches as a map or graph, which could improve navigation.

Eventually, we can envision a ChatGPT equipped with a full tree view, allowing users to merge branches, associate metadata with them, and even collaborate on them with others.

The introduction of the “Create Branch” button is not merely a user-friendly enhancement. It reflects a deeper ambition: to transform the relationship with AI into a structured, iterative, and evolving process. This change in the interface accompanies a conceptual shift: the user is no longer simply engaging in a dialogue, but participating in an assisted co-creation process, where the paths explored have a history and a form.

As AI interfaces evolve beyond simple chatbots to become true cognitive environments, this feature could well become one of the cornerstones of a future use of artificial intelligence that is reflective, collaborative, and creative.

On this blog, we’ve already explored ChatGPT’s evolution toward advanced features. Check out the following:

1. OpenAI. (2024). ChatGPT Update Notes – Introducing Branching for Conversations.
https://help.openai.com/en/articles/8985934

2. TechCrunch. (2024). ChatGPT adds “Create a Branch” button to improve conversational navigation.
https://techcrunch.com/chatgpt-branch-feature

3. OpenAI. (2024). ChatGPT Help Center – How tools and custom instructions work.
https://help.openai.com/en/articles/8821930

Don't miss our upcoming articles!

Get the latest articles written by aivancity experts and professors delivered straight to your inbox.

We don't send spam! Please see our privacy policy for more information.

Don't miss our upcoming articles!

Get the latest articles written by aivancity experts and professors delivered straight to your inbox.

We don't send spam! Please see our privacy policy for more information.

Related posts
Generative AI

DLSS 5: Nvidia promises photorealism powered by AI, but questions remain

Real-time graphics are entering a new phase. For several years now, the video game industry has been striving to make digital imagery more photorealistic by combining computing power with software innovations. With the announcement…
Technological Advances in AIGenerative AI

Anthropic Enhances Claude: Toward an AI Capable of Creating Interactive Visualizations in Real Time

Conversational artificial intelligence is entering a new phase of maturity. Until now, assistants like Claude, ChatGPT, and Gemini have primarily stood out for their ability to generate text, code, or complex summaries. But…
Generative AI

OpenAI unveils GPT-5.4, a model designed for complex reasoning and coding

GPT-5.4 is available in two main versions: GPT-5.4 Thinking and GPT-5.4 Pro. Both versions are based on the same architecture but differ in terms of performance, speed, and pricing. One of the advancements…
The AI Clinic

Would you like to submit a project to the AI Clinic and work with our students?

Leave a comment

Your email address will not be published. Required fields are marked with *