properbrew

> to do great things in the context of the right orchestration and division of effort

I think this has always been the case. People regularly do not believe that I built and released an (albeit basic, check the release date - https://play.google.com/store/apps/details?id=com.blazingban...) android app using GPT3.5. What took me a week or two of wrangling and orchestrating the LLM and picking and choosing what to specifically work on can now be done in a single prompt to codex telling it to use subagents and worktrees.

> We define an MAU as an authenticated Pinterest user who visits our website, opens our mobile application or interacts with Pinterest through one of our browser or site extensions, such as the Save button, at least once during the 30-day period ending on the date of measurement.

Wonder if we're going to get a MAHU (Monthly Active Human Users) stat in the future.

I meant simple in a good way.

I'm utilising codex within an application for some simple inference stuff in read-only mode, but this gives me a way to easily wrap it in some guardrails.

Looks simple but very useful. This came across my screen at just the right time, have a few ideas that could do with something like this.

I don't think it's a case that no one liked it, there's just too much going on that it probably never came across the right eyes.

I linked one of my projects in a post and it got some really good responses. I did a bit more work on it and posted a Show HN thinking a few people might be interested but it got 0 traction.

I even made it a point to go on the new Show HN and checkout some peoples projects (how can I expect anyone to check mine out if I'm not doing the same) and it is hard to keep up.

I have another app that I've been working on for the past 3 months and whilst I want to do a Show HN to discuss how I built it, the moments I was banging my head on the wall working on a bug etc, I sadly wonder if there's any point.

If you're looking for free STT you can use Whistle across Windows/Mac/Linux and Android (iOS released soon)

https://blazingbanana.com/work/whistle

If you're looking for free STT you can use Whistle across Windows/Mac/Linux and Android (iOS released soon)

https://blazingbanana.com/work/whistle

I was using a custom skill to spawn subagents, but it looks like the `/experimental` feature in codex-cli has the SubAgent setting (https://github.com/openai/codex/issues/2604#issuecomment-387...)

Thanks for the reply, will pass this info on.

> Actually it may just be aimed at investors

The fact that the first image you see has "$60M seed" in big text, I have to agree, this does not feel aimed at devs.

Do you have any links for how you did these stretches? I know someone that could do with trying this.

You also don't have to throw away everything you've learnt in those 8 months, there's some things that you'll subtly pickup that you can carry over into the next generation as well.

Thank god it's only Actions, Copilot, Issues, Git, Pages, Packages and Pull Requests....

Now it seems Actions has broken - https://www.githubstatus.com/incidents/lcw3tg2f6zsd

Thank you for this, this was a really interesting read about batch invariance, something I didn't even know about.

> I want to see some assurance we get the same results every time

Genuine question, but why not set the temperature to 0? I do this for non-code related inference when I want the same response to a prompt each time.

SQLite is awesome, no services to run, ports to open or anything like that, just a nice little database to use as you need.

If you read the article you'll see it's not about generating AI slop images:

> Take a freeform query (like ‘sfo->jfk’) and turn it into a ‘place’

> Build a database of ‘places’ -> pictures

> Build a software system that can take a ‘place’, look it up in a database and spit out the right picture – even if that ‘place’ isn’t in the database

These were great reads, thanks for linking. The writeup around the UUID page was super interesting!

For writing decent code, absolutely not, maybe a simple bash script or the obscure flags to a command that I only need to run once and couldn't be bothered to google or look through the man page etc. I'm using smaller models for less coding related stuff.

Thankfully OpenAI hasn't blocked me yet and I can still use Codex CLI. I don't think you're ever going to see that level of power locally (I very much hope to be wrong about that). I will move over to using a cloud provider with a large gpt-oss model or whatever is the current leader at the time if/when my OpenAI account gets blocked for no reason.

The M-series chips in Macs are crazy, if you have the available memory you can do some cool things with some models, just don't be expecting to one shot a complete web app etc.

I didn't even get to send 1 prompt to Claude and my "account has been disabled after an automatic review of your recent activities" back in 2024, still blocked.

Even filled in the appeal form, never got anything back.

Still to this day don't know why I was banned, have never been able to use any Claude stuff. It's a big reason I'm a fan of local LLMs. They'll never be SOTA level, but at least they'll keep chugging along.

Not even an Apple fan, but from what I've been testing with for my dev use case (only up to 14b) it absolutely rocks for general models.

> you can get recommendations from artists that belong to playlists you follow or you've created

Does this include the "Liked Songs" playlist? I'm terrible at actually following artists.

Didn't know this was a thing I needed. Love it!

Still working away on a free, completely offline transcription app that's available on all platforms with CUDA support as well.

https://blazingbanana.com/work/whistle

Currently tidying up some internal code (also removing the larger model on mobile platforms) and implementing proper diarization (who said what) so that it can be used for more than just personal dictation.

My iOS developer account is _finally_ approved so it will be available through the proper app store soon.

I think it depends on how determined the actor is. I see all the range from your simple scripts to full on mimicking real user behavior that I can only really spot from the honeypots they hit.

You'd probably catch most the low hanging fruit for sure, but you would cause friction for real users.

I say this as someone who has enabled captcha on some of our more critical endpoints, there's definitely a place for it.

It may be simplistic, but that's a cracking photo you've taken on it.

Reddit post in question - https://old.reddit.com/r/confession/comments/1q1mzej/im_a_de...

https://archive.ph/1XUka

> A company spokesman told Podnews that publishers did not opt-in to be included, and “there is no opt-out mechanism at this time”. The company also clarified that they do not compensate publishers directly.

Well that seems like a really shitty thing to do. I'm all against targeted ads, but creators need to have a way of earning money. I know we've moved to an "ask for forgiveness rather than permission world" but they could have even reserved some amount ready to pay.

The name rings a bell but I hadn't looked in a long while. The videobridge might be very useful when I look to expand from just 1-2-1 video calls, so thanks for the nudge in that direction.

I know it's stuff that's been done before, but it's fun seeing how this stuff works and fits together, along with the limitations you have within a browser environment.