Hacker News: mtaras

New comment by mtaras in "The UI future is colourful and dimensional"

mtaras — Tue, 27 May 2025 09:23:53 +0000

I hate those Airbnb icons. They look like nothing, just objects in a somewhat cartoon-y style. I much prefer where both Apple and Google move UI design.

New comment by mtaras in "The unreasonable effectiveness of an LLM agent loop with tool use"

mtaras — Fri, 16 May 2025 14:27:24 +0000

"Message me" is through a bot in my main messenger app, so the agent is just a chat window for me. "Later" happens per my requests, and agent has an ability to do something later because I just automatically invoke it on a schedule periodically, and ask it "check notes and chat history — do you need to do something, like message the user to remind them of something", and agent has the message_user tool to do that (that goes to the messenger bot API). Or it can decide to do nothing.

New comment by mtaras in "The unreasonable effectiveness of an LLM agent loop with tool use"

mtaras — Fri, 16 May 2025 12:16:23 +0000

No, I personally don't have notes that would be valuable enough for the system I was looking for, so notes is just how I'm calling my agent's long-term memory. All I do is provide it with tools to message me and manage it's own notes, useful context (recent chat history, its saved notes, calendar events, etc), and it can act upon the info.

The elegance of the system unfolded when I realized that I can not specify any interaction rules beforehand — I just talk to the system, it saves notes for itself, and later acts upon them. I've only started testing it, but so far it's been working as intended.

New comment by mtaras in "The unreasonable effectiveness of an LLM agent loop with tool use"

mtaras — Fri, 16 May 2025 07:33:55 +0000

Just a couple of days ago I discovered this truth myself while building a proactive personal assistant. It boiled down to just giving it access to managing notes and messaging me, and calling it periodically with chat history and it's notes provided. It's surprisingly intelligent and helpful, even though I'm using model that's far from being SOTA (Gemini Flash 2.5)

New comment by mtaras in "Gemini 2.5 Pro vs. Claude 3.7 Sonnet: Coding Comparison"

mtaras — Mon, 31 Mar 2025 17:06:56 +0000

("it" being the Claude 3.7, not the Gemini)

New comment by mtaras in "Ggwave: Tiny Data-over-Sound Library"

mtaras — Tue, 25 Feb 2025 22:39:15 +0000

This sounds delightful, I might make esp32s talk to each other like that just because it's adorable

New comment by mtaras in "Grok3 Launch [video]"

mtaras — Tue, 18 Feb 2025 09:00:30 +0000

The Vergecast recently did a section where they asked listeners what they use LLMs for (specifically not for coding) https://youtu.be/WwNjBNtZ3Co 30 minutes starting at 45:25, it had a number of interesting examples. Might not convince you of LLM's excellence, or might not be much different from what other people commented, but it's a good listen nonetheless.

New comment by mtaras in "Gemini 2.0 is now available to everyone"

mtaras — Wed, 05 Feb 2025 17:03:36 +0000

Updates for Gemini models will always be exciting to me because of how generous free API tier is, I barely run into limits for personal use. Huge context window is a huge advantage for use in personal projects, too

New comment by mtaras in "E Ink faces growing competition in the "paper-like" display space"

mtaras — Sun, 01 Sep 2024 10:31:03 +0000

MKBHD did a review of a TCL phone with an NXTPAPER display and the screen looks like just a normal LCD with a matte screen protector. It has some extra layers physically to limit the amount of blue light, but it 100% looked like any other phone screen with a "paper-like" screen protector on top. The only other thing were the screen modes that would wash out colors (partially or completely) to imitate the "paper look", but it's still pretty much glowing like any other regular display