Joplin is a FOSS, multiplatform, Nextcloud-compatible note-taking app. Apparently, the dev uses Coderabbit and is now pulling integrated chatbot support from the optional AI plugin to the main (dev) branch. (Looks like Joplin Cloud, their note syncing and sharing service, has had chatbot integration for a year, but the user needed to opt into Joplin Cloud and the AI plugin was required.) At least the functionality is disabled by default but that might easily change later...
Screenshot of https://github.com/laurent22/joplin/pull/15643
Chill out, he added a function for plugins to implement ai in a consistent way. Nothing wrong here
Not a bad idea to make infrastructure separating the model-calling backend from user-facing frontend: only a single plugin manages local models and API key to remote models.
But isn't this a bit of scope creep for a note-taking app? As I understand it, this whole AI framework has existed as a first-party plugin for a while now, with no AI API in the base app (which is quite bloated as is).
Not to mention, lots of recent contributions are heavily AI-made: not ideal for an app handling important private stuff.
I think so too. It seems the change solely benefits llm users, and that support for llm integration would be more or less the responsibility of the maintainers from here on. If chatbots change or expand their api, auth mechanisms, add new features that can't go through this plugin interface etc, it would continue to divert core maintainer work on behalf of llm plugin maintainers, and increasingly tie the project development to llm development.
From what I've seen over the past few weeks/months, more llm users are agentizing and harnessing and looping and whatever and 'calling out' more, and it seems that interfacing with llms is only going to become more complex as the base models continue to provide diminishing returns and less stable outputs for their increasing power and time usage.
Not to mention, Joplin allows more than structured text (native embedded images/media, even more kinds of content with plugins) and the text-only
joplin.ai.chat()will definitely "need" to be supplemented with multi-modal AI support.As for new techniques, I imagine the declining AI subsidies will raise the popularity of context reuse (reducing token use by not having the LLM re-read the whole notebook; is that what it's called? I don't remember much from videos/articles about AI trends because I don't use it). This will require a huge framework for managing data structures distilling every cached context window... And I think non-LLM functions will be added too to save tokens, for example an LLM may want to know "is there an unfinished TODO?" without reading all of them.