Introducing VizThinker: A New Way for Visual Thinkers to Interact with AI

======================================

Traditional Chat Box flow and Visual Thinking Person

When I use LLM chat boxes like ChatGPT, Gemini, and Grok, I find them extremely helpful for enhancing my work efficiency and productivity. Specifically, they excel at data collection. I do not need to search through web pages to teach myself about a new topic. For example, if I have a confusion about CI/CD tags and ask an LLM to explain them to me, it can give me a response that lists a comprehensive explanation, including the concept behind CI/CD, the most used tags, and some examples. It is definitely a great response but also lengthy. When you want to explore several branches from this response, such as asking about the general use cases of CI/CD, it will give you another long response. At that point, if you want to look back at the initial response, you will need to scroll back, which takes your time and energy.

Plus, I deem myself a “visual thinker” and constantly feel the traditional format of a chat box, from top to bottom, is not the best way to aid me in composing my ideas and learning by leveraging the power of LLMs. These reasons ignited my idea for VizThinker.

VizThinker AI

VizThinker AI lets you interact with an LLM through a node-based interface where every prompt, response, or follow-up becomes a node in a graph. Dig deeper into a topic (vertical nodes) or branch off to explore related ideas (horizontal nodes), all while seeing your thought process come to life as a visual map.

  • Node-Based Chats

    • Start with a root prompt, then add follow-up or branching nodes to dive deep or explore tangents.

  • Dynamic Graph

    • Watch your conversation grow into an interactive graph with clear vertical (deep dive) and horizontal (side quest) connections.

  • LLM Power

    • Send node content to an LLM (like OpenAI) and get responses as new nodes.

  • Interactive Canvas

    • Pan and zoom to navigate your idea graph.

  • Save & Load

    • Store your sessions locally on your desktop and pick up where you left off.

  • Conclusion

    • Consolidate all connections of nodes to generate a holistic report based on your idea graph.

Future Work

Basically, this project is still in its very early stages. I will keep working on it and update the progress here. Are you a visual thinker? If yes, please let me know. I’m really happy to listen to your experience when using LLM chat boxes and share your ideas for VizThinker.