Hello folks,

How do you use outliners together with AI (if you do)? Do you use outliners less often? Which new patterns do you notice?

We've recently added a MCP server to Checkvist, which allows to read/edit your lists with LLMs, but I'm yet to understand how useful it is.

Personally, I sometimes use LLMs to summarize the data in some big lists or ask for insights, but I don't do this often, so my habits of working with data haven't changed much.

Sorry for the open question :)
KIR

I don't use AI to compose. I'll ask Claude to "give me a 150-word explainer about koala bears", but only use it as a side note, not in my own composition. I allow macOS to check spelling, but never ever use computers for grammatical or style check. Either my work meets the bill, or it doesn't, but I learn more from readers and clients about what works than from robots.

I do however enjoy having long conversations with Claude about science or other topics that interest me. It doesn't matter if Claude is telling me the truth or not, since none of what we talk about affects my work or personal life.

Despite doing a lot of experimental work with LLMs (much like Paul), I don't use them for serious work because (a) I don't trust their accuracy, (b) I don't like their constant efforts to persuade me that they're the equivalent of a human colleague, and (c) due to their environmental impact (although read Ed Zitron on the reality of AI investment – it makes you despair of humanity, honestly). I do use a local model as a kind of super-thesaurus if I'm feeling uninspired, but I certainly don't allow it to dictate my style.

Friends in academia despair of ever receiving original compositions again in their lifetimes. If award-winning novelists can (allegedly) get by without actually writing anything original, then where can we end up?

In retirement now, I intend to start writing book-length pieces, and I do not intend to use AI directly. Seems impossible to avoid AI altogether; I mean search engines are using it, right? But if I want to read something not written by me, I'll buy a book -- and hope it is original work.

No doubt you'll keep us in mind when early-access editions are published? :-)

Stephen Zeoli wrote:

In retirement now, I intend to start writing book-length pieces

Thank you, everyone, for the answers. From what I heard, working with AI is mostly separate from the actual work you do, so integrations between LLMs and the actual tooling are not a hot topic (yet).

Personally, I sometimes ask AI to summarize my outline with daily notes (to get an overview of what I was working on the previous month), but again, this is a rare use case.

Thanks again :)

After upgrading to macOS 27 several AppleScripts I had in Typinator stopped working, because they depended on an obsolete JSON parser that Golden Gate refused to run. I gave them to Claude which rewrote the scripts in seconds and they now work again in Typinator. It also explained why it did what it did and made a couple of tweaky suggestions that were useful. This, for me, is the value of AI. Quickcoding things I use just for myself, and no longer want to struggle with AppleScript to figure out.