even though it's a builders' decade, it's still a challenge to build something more complicated than a to-do list

here's an example

knowledge is easier to comprehend when it's structured like a tree. you have a starting, broad point (the root), and then you start deconstructing each part until you get to the fundamentals (the leaves), which become self-evident

this kind of deep dive is unimaginable now without LLMs. you may have already felt that research is the most reliable use case for non-tech people—even if you still need to double-check for hallucinations, they happen much less than they did a year ago

however, chat is not an ideal UI for this:

• it represents only one 'branch' of a tree at a time
• you have to scroll all the way to the top to go deep on another part of the initial statement
• you can't see all the branches at once—and visuals are powerful for human memory

a mind map would work, but how do you make one with an LLM?

• i don't want another tool for it. i want this in my everyday AI chat, with my history, API keys, deep research features, MCP servers, and everything else i use
• if you're using the default ChatGPT or Claude apps, you can forget about it and wait till they release something like this (=never)
• it would take months to build the same functionality from scratch (and a lot of wasted energy generating code)

i use TypingMind for my daily AI (i really love it!), and luckily it's quite extensible. you can build features on top of it (i already built one, as i mentioned here https://lnkd.in/dZvnQwPS)

but then you suddenly become an engineer to solve this not-so-simple task:

• you have a platform to extend, with its own codebase and specifics you need to understand.
• you have to choose a canvas technology—which one is feasible to integrate? easy to implement for this use case?
• and you have to tie those two together: how to inject it into the existing messaging and branching logic? how to embed a new view?

doesn't sound like something Lovable could solve, right? i'm not saying it's impossible, but it requires much more investment than just an evening or two to get it working

i wish for much more extensible software, where apps are compliant and integrated from day one, because non-integration simply isn't an option. it feels inevitable, but it's a whole another story