mistercheese

I agree, most of the proactive agent assistance use cases requires me to give out data I’m hesitant to give like email, financial data, health data…

I’m hoping someone makes a lower maintenance version of open source Muse or Grokbot or Dots (more user friendly than OpenClaw or Hermes) that I can easily deploy to my own AWS, Cloudflare, Vercel, or Fly account and use some ZDR friendly inference provider like Fireworks.

Yeah I actually don’t even use Paseo’s orchestration much. I’m mostly using Paseo because I want to use Anthropic and OpenAI’s native harnesses for their respective models, but OpenCode for others, with a good remote mobile app experience when I want to check in on agent work or steer things away from my computer.

I haven’t found a better OpenCode remote mobile app experience. If OpenCode or Pi makes one, I might just move to that.

I’m not sure if I missed it, but I’m curious how you feel about cloud hosted models with ZDR policies? GLM5.2 or even Minimax M3 on Fireworks or Together ai should be still relatively/consistently cheap and private but a lot more capable and easier to setup?

I thought only the MmiMo 2.5 non-pro was multimodal?

Do you have a harness recommendation? Sounds like maybe you’re into Hermes?

I have this exact same fear as an IC.

I wonder if Engineering Managers have this same fear, or they’re used to having to distribute complex tasks to senior engineers and gamble with seeming less risky tasks to juniors that may leave ticking time bombs in their code. Just the nature of code written by agents or humans?

This is a great idea! I’d be curious to learn more about your setup, particularly how it asks your family and follows up with further questions? Does it do it based on a graph of your family it builds real time? Or is it smart enough that you just prompt it to “follow up with more questions”? I’m having a hard time imagining it asking my family engaging questions they won’t just say, “I don’t remember”.

That’s a good point. I think I saw Together.ai with that offering, but for some reason just never think to throw random non urgent coding tasks at it overnight

Wow what kind of hardware do you have? Mac Studio, dgx spark, strix halo? How fast is it?

I use local models for asking about personal financial or health data that I want to keep local and private. Or even just whipping up quick and dirty prototypes for whatever I can think of but not seriously enough to spend tokens that I rather use on real projects.

Yeah I think there’s benefits to third-party providers being able to run the large models and have stronger guarantees about ZDR and knowing where they are hosted! So Open Weights for even the large models we can’t personally serve on our laptops is still useful.

Nice! With Anthropic and OpenAI lowering usage limits that’s a big deal. I’ve been considering a Gemini subscription because I’m already paying for the Google storage plan and it would only be $10 a month more for me.

Do you use Gemini 3.1 pro or the flash model or just auto? Any feelings about quality vs other models?

Out of curiosity, why do you like gemini-cli better than claude or codex? And do you have any comparisons to opencode or pi?

Personally I've really liked OpenCode's TUI, but maybe on a superficial level of "this looks good and feels ergonomic to me".

Gemini cli felt clunky for me when I tried a while ago, but maybe it's better now? I do like how it's open source and I'm wondering if it can be made as model agnostic as OpenCode.

Is it feasible to run LLM inference comparably without CUDA or Rocm? How much of the cost performance goes away?

Would love a zdr option if possible, that’s honestly the main thing I’m going to OpenRouter for.

Yes I thinks somehow we need a bulldog check gate before it even goes to a human reviewer

> the most important duty of the submitter is to vouch for it

When shipping pressure comes, I’ve seen this to be the first thing to go. Despite formalizing ownership standards, etc… people on both the submitting and reviewing end just give up understanding Ai slop when management says they need to hit a deadline.

Probably no company would actually do this, but I wonder if we should actually actively test the submitter’s understanding of the code submitted somehow as a prerequisite to moving a PR to ready for review. I’m not sure if it will be actually hopeful, enforcing people to understand the code, but maybe at least we’ll put the cultural expectation upfront and center?

thoughtful article

How did you find your mix of freelance work initially? Do you go through agencies? If so, how did you find the best agencies starting out?

As a complete newbie to contract/freelance work, how did you first get started finding good agencies to work with?

Maybe I'm no fun, but am I the only one who fails to see why this is at all better than a normal, boring Connect Four App? This isn't the correct side board, and for the effort of attaching and un-attaching the case, it only adds a restrictive, small, physical grid which could just as easily have been drawn onto the screen anyway.

Forgive my crowd funding ignorance, but what's to stop them from just paying themselves (as a mysterious donor) whatever the gap is at the end so that they keep whatever amount they do manage to raise?

I'm curious if the ability to pilot the device is that useful - it can't navigate crowds or do stairs anyway. So for the price, an iPad on a stick that has any always on VC would seem to be much cheaper and just as effective. Easy to move around if you need, and cheap enough to put in every room.

I would wager that even so, drives are replaced in less than 10% of all laptops bought by consumers. Even batteries would likely be a small portion.

Now if we talk about this particular upgrade, how many people would pay $600 bucks for 960 GB of space, I can imagine it's an even smaller fraction.

The size benefit is something that everyone appreciates when they buy a MBP retina. There are enough other laptop choices if you want upgrade-ability and repair-ability, and MBP retina doesn't need to compete on that level.

This looks awesome, I've been curious to how we might be able to programatically solve for "surfacing interesting content". Upvote solutions are on a decline, and curation/moderation is on the rise as we realize those solutions don't scale all that well. Apply to music, videos, photos, links, comments, etc...

Or in the new tab window in the bottom right there is "Recently Closed" where you can unclose tabs and groups of tabs.

Except this isn't a crash, it's something that Google is explicitly doing on their end from the reports.

I too can't help but feel like I'm missing some context, seems like Google would at least warn the user.

I don't understand how Firefox/Chrome getting rid of their in-navbar subscribe or Google not advertising Reader doesn't just put them at the same level of Twitter when it was just starting out. Twitter didn't necessarily need large marketing efforts to take off as far as I know.

I feel like this is a case of the tech savy being somewhat removed from the majority of the web's consumer base. Honestly, I've had a surprisingly large amount of trouble explaining something like IRC/RSS to someone like my parents or my non-techy friends. I try to explain that IRC is just like any other chat room, but honestly it's incredibly scary and unattractive to most people.

For HackerNews readers, and also for me that's probably true. However, the popularity of Twitter would indicate to me otherwise.

I would argue RSS was neglected by Twitter doing certain things better, things that the average user cares about