Pretty cool! The branches i got back from my prompt felt a bit too concrete, though. My prompt probably sucked, but sub-node felt .. off. It just wrapped my prompt.
Love the idea though, will explore again in the future
for what it's worth as someone who uses mind mapping tools daily for brainstorming this seemed pretty intuitive to me. I was up and running mapping something out pretty quickly. I wasn't sure what the brain and detective icon were so I just clicked them and found out
This is really cool and quite fun to use! I very much enjoy having all the information on one screen and pair with an LLM to brainstorm and test ideas.
I'm wondering if there is any way of running this locally, against my own models?
I understand your frustration and and sorry for the lack of clarity. Let me provide you with a brief explanation of what this is all about.
NodePad is really a simple experiment. The purpose of is to tackle the issue of inconsistent responses in Language Models (LLMs) and explore potential solutions through the user interface (UI).(hence the video, I understand though that it wasn't exactly helpful :) )
The tool itself is quite straightforward. It allows you to express your thoughts rapidly by typing them into the input field at the top. Once you've entered your thoughts, you can then further develop or question them individually using the brainstorming feature. You may connect them visually or just download the cancas as a markdown file.
You really need to work on concisely getting to the point. There’s a good idea here.
I think you want to say something like “NodePad is an AI brainstorming partner. It can contribute to ideation by contributing ‘nodes’ to a thought-cloud, that develop or question your ideas, helping you explore and clarify your thinking”
I think you’re very interested in why you think the chat interface to LLMs is not great, and why you think your model might be better. But you don’t need people to follow that chain of reasoning for them to see the place you ended up as being interesting.
Show, don’t tell. And start with what the tool does, before you explain why it does it that way.
Might this conversation be an example of the type of thing that you are ultimately trying to address here? Like, could it qualify as a valid use case for using such a tool (if people were serious enough about the topic)?
It seems to me that intuition plays a heavy role here.
When did you start this tool before ChatGPT3/4, or after?
Feels like it was started before LLMs, and now this pitch is just trying to incorporate them after the fact, as opposed to rethinking your thing from a deeper level, now that the world has changed.
Seems like you have a unique skillset in having some mental machinery to view the world in your 2D grid/node based paradigm. If text brainstorming was your use case, I think ChatGPT has killed that. If I'm gonna visually brainstorm I'm gonna use pen and paper, and if I'm going to be on my computer doing text brainstorming I'm gonna use ChatGPT. I don't think the lines provide enough value (chat GPT does bullet points well and can also generate graph viz)
Is there some end-product your tool could produce that users want, and LLMs can be used to help make those? Or can you take more advantage of your 2D + grid/node based tool to provide something more than an obscure text editing experience? For example, perhaps using mathematical distance as a first class citizen, adding physics, et cetera (I have no idea what would be useful, just that perhaps there are some "magical" things that a node tool could do that a text/chat interface will never come close to)
Your original Thoughtpad.space seemed way simpler for laptop trackpad usage and could easily double click to delete and single click to reorient the items etc. why the change?
Heya, the grid had some limitations as I was re-evaluating some features. Tried to keep things simple here as well. Where did you find the complexity in the new one on a laptop?
without watching the video that sounds .... not straightforward. It sounds like you've dupliacted a chatGPT interface and then... i dunno added a way to draw lines between answers?
The elevator pitch for this really needs work. Like Maybe a simple screenshot of the result + a description of the bits that aren't obvious from that would be helpful on the site.
hopefully constructive suggestions:
> NodePad is a simple LLM-assisted brainstorming experiment designed to help you quickly capture ideas, expand on them, question them, and organize them visually!
a) don't care if it's LLM assisted. and only geeks know what LLM means. I care if it's going to intelligently help me
b) "x is a brainstorming experiment" means that x is a brainstorming experiment NOT that x is an experiment that helps you brainstorm. These are very different things. I think you mean "NotePad is an experiment that uses AI to help you brainstorm"
c) organize them visually could mean literally anything with a visual component. You gotta give me more words or a screenshot.
> without watching the video that sounds .... not straightforward. It sounds like you've dupliacted a chatGPT interface and then... i dunno added a way to draw lines between answers?
IDK, if someone did just that, I'd happily pay for it - especially if those answers and lines formed a DAG, offering a nice interface to fork and merge conversations.
have you considered... considering the context of my response?
the discussion i was responding to was about the contents of the page PRIOR to watching the video so the comments were made as feedback based on what a reader sees BEFORE watching the video.
Watching the video would have
a) polluted the honest feedback about what a user saw when encountering the page and trying to decide if watching the video was worth the time
b) been !@@#$!@ irrelevant to the discussion of what one thinks BEFORE watching the video.
Thanks, agreed makes a lot more sense to present it this way. I think I was trying to explain a lot of ideas I had while working on this to a point where things became either redundant of not helpful. I'll adjust the about section.
It appears to mark anything remotely related to "destruction" as restricted content. Or rather, you can clean a phrase by using the word "saint".
Won't work: How should a demolitions expert expect to be treated by an architect?
Works: How should a saint expect to be treated by an architect?
(silly questions I know, just to make the point)
Why would a text tool have any content restrictions? It's ridiculous. Microsoft Word, Google Docs, Notepad, vim, etc. do not prevent me from writing about anything.
If I want to mind-map political joke ideas or brainstorm adult toy names that's my business.
I can't even use this tool to brainstorm about the pros and cons of different content restriction systems, the irony.
https://github.com/qrdlgit/simbiotico