Harbor Bai   
   
          Selected Works             About                                 CV

Dive-in LLM

A spatial GPT workspace for immersive idea exploration.





  • Role: 
    Web developer
    UX designer
  • When:
    Nov 2025
    1 week solo project
  • Focus: 
    AI/human collaboration
    UI/UX
     
  • Tools:
    Replit
    OpenAI API 
    HTML
    CSS


  • Check out the live web app! 
    Overview: Dive-In LLM reimagines the way we interact with AI by replacing linear chat with a spatial, cinematic workspace. Through “screen diving,” users can explore sub-conversations in layered panels — preserving context, hierarchy, and flow. 
    Build with React and Framer Motion, the system transforms dialogue into a navigable landscape of ideas, enabling deeper focus, seamless branching, and a more intuitive form of human–AI collaboration.







    Painpoint:
    Current chat interfaces cause cognitive overload — linear threads make it hard to recall context or compare related ideas, forcing fragmented focus as users dive into subtopics and lose sight of the main thread.

    Without spatial memory, text scrolls endlessly while the mind craves structure, hierarchy, and a sense of progression. As a result, creative exploration feels constrained rather than cinematic; users can only read through ideas instead of moving through them.



    Screen diving:




    Solution:

    A spatial, layered GPT workspace that transforms chat into an immersive, explorable environment.

    Through “screen diving”, users can open sub-conversations in parallel panels — each inherits context while preserving hierarchy. This design replaces endless scrolling with navigable depth, enabling users to branch, compare, and synthesize seamlessly.




    Why it works:

    A conversation is never a linear thinking:



    Designers, researchers, and thinkers operate in layers — zooming in, comparing, mapping relationships. By visualizing dialogue spatially, Dive-In LLM aligns the medium of conversation with the architecture of thought.

    It’s not just about better UX — it’s about redefining interaction between humans and large language models as an act of spatial reasoning.



    Design process:
    The design process evolved from identifying the limits of linear chat to building a cinematic, panel-based system—mapping conversations as hierarchies, prototyping spatial interactions, refining visual and emotional flow to make dialogue feel alive and navigable.





    Conclusion :
    In the end, Dive-In LLM is more than a new interface — it’s a new way of thinking with AI. By transforming dialogue into space, it turns conversation into exploration, and reflection into design. It reminds us that intelligence isn’t just what we say — it’s how we move through ideas.






    Harbor Bai © 2025. All Rights Reserved