I often use ChatGPT for long technical and research conversations.
A recurring problem is that I may understand the overall answer, but not one specific concept inside it. I then ask several follow-up questions about that concept.
This quickly pushes the original conversation far upward and makes the main discussion difficult to follow.
The existing Branch in new chat feature partially solves this, but it creates another full conversation every time. For small clarification questions, this can generate many unnecessary branches.
I would like an inline side-thread / nested-thread feature.
For example:
Main conversation:
We can use Test-Time Training to adapt the model…
I don’t understand “adapt the model”, so I click something like:
↳ Ask about this
A temporary side panel opens:
-
What does model adaptation mean?
-
Does it require backpropagation?
-
What parameters are updated?
-
How is this different from fine-tuning?
After I understand it, I close the panel and return to exactly where I was in the main conversation.
The main conversation would only show a small marker such as:
↳
Model adaptation — 6 messages
I could click it later to reopen the discussion.
Ideally, the side thread could also generate a one-sentence summary that becomes part of the main context, while the full clarification conversation stays collapsed.
Why this would help
Long ChatGPT conversations are often not truly linear.
They are more like:
Main task
→ temporary clarification
→ clarification inside that clarification
→ return
→ continue main task
This is similar to interrupt handling in an operating system: pause the current execution flow, handle a temporary issue, then return to the original point.
Branching is useful when I want to explore a different direction.
Side threads would be useful when I simply want to understand something before continuing in the same direction.
For research, learning, coding, and other long conversations, I think this would make ChatGPT much easier to navigate.