You're not using LLMs enough

A consistent theme I have observed lately by working and talking with a lot of different people is that a lot of them are underestimating the capabilities of LLMs or they don't know that things can be easily automated these days.

They are mostly stuck in their old habits and procedures and use a lot of their valuable time for things they should never be doing. They are not using LLMs enough and you probably aren't, either.

I really like this XKCD:

Because it shifted so much in the age of LLMs, as the real time invested dropped substantially^[1]. Nowadays, you can specify the task you want to automate, fire off Codex or Claude Code, check back in 30 mins and have an almost-perfect result.

To give you some ideas, some things I have automated or do regularly using LLMs:

• Convert images or timetables into calendar entries by uploading the image and asking for a .ics file.
• Wrote a bunch of scrapers to check for the existence of something or to collect data.
• Have scripts that automatically translate the clipboard into another language by using Espanso, one of my favorite tools.
• Use Obsidian Clipper to transform certain web pages into semi-structured markdown files by using the Interpreter and to extract key information.
• Keep different calendars in sync by using a small Swift script and macOS LaunchAgent. This setup alone has saved me a $50/yr SaaS subscription and a lot of (rather useless, but necessary) copying.
• Syncing a set of files, but only starting the sync when connected to a VPN.

A lot of these things (probably all of them) don't matter to you and your life in any way, so I spare you the details of their setups (although some of them are quite impressive). But that is exactly the point: We are in the age of truly specialized software and for just 20 bucks a month^[2] you can unlock this power and spend your time on more important things: Learning new things, writing or doing important work.