Where is your ideal customer? What do they read and watch? If you can answer those questions, you've likely found your answer. If you can't answer those questions, then identifying who your ideal customer is and where they spend their attention is where you should focus.

Views per video aren't as important as sales per video (or leads per video).

Which type of content probably depends again on what you're ideal customer knows and doesn't know and the problem they're trying to solve right before they buy. (If you sell $20k-500k solar systems, it's different than if you sell plushy animals.)

Sorry to be vague, but a general question can only get a general answer.

(If reels are working, man, keep it up. More sales means more money to experiment with different content and platforms. LTV:CAC is one of the most important numbers you can know. Try to get LTV up so you can afford customers at a higher CAC and you'll have freedom to try, fail, and learn in your marketing.)

Okay thanks, I will keep it in my mind , and work according to this

Love to see the World English Bible here. (Nothing against the ASV or BSB, I'm just used to seeing KJV and ASV as the only options when someone reaches for a public domain translation.)

The WEB is woefully under-utilized for such a good, readable, modern and openly licensed translation. I do read from the Greek New Testament and when I've compared the WEB with it, I've been pleasantly surprised at the level of Greek nuance they were able to capture with their English phrase choices. (I do not know Hebrew and have not checked the WEB on that.)

Thanks, I’ve enjoyed it as well though I’m typically using ESV and NKJV. I found out about WEB from an app my wife uses for listening to the Bible, the recordings are also openly licensed and audio is something I’d like to add to the app in the short term.

What are you using your local LLM for? What's the smallest/weakest model (and probably at what speed) that would make the hardware spend worth it?

having your own hardware lets you do much more than serve LLMs, it's a great learning platform in general

the other thing is that the little models are getting better and more capable, so the hardware ROI becomes greater over time

Ideally I want to even train or at least fine tune, but I can’t afford a Blackwell.

I'd recommend to start by using a platform for first experiments, like fine-tuning offerings, which can be done for lower cost and will handle the thornier issues therein.

Do some experiments to see if fine-tuning shows potential for improvements, curating the dataset and reward function are the harder parts anyway.

I’d like coding assistance and light logic.

Maybe the DeepSeek R1 distilled

If you did this well, you could organize the world's information.

It might even make it searchable.

And I'm sooo thankful we don't. As a guy who tries to save money by DIYing it in 3x-10x the time it should take, I would be broke.

We all use screenshots so much. They are the [universal API](https://mattrickard.com/screenshots-as-the-universal-api)

I am not surprised by so many people wanting personalized solutions. So many screen shot apps and annotators are so close, but just not quite right for a person's workflow.

I'm using off the shelf open source solutions now, but even those I have had to modify/patch to add just a couple of things to make it smooth in my workflow.

What value to you get from having this happen on every turn with a change? I get that the ratio should be a lot higher than the default. Doesn't doing this for every turn where something changed slow things down a lot without gain? I haven't experimented, that's why I'm curious. Why not 1 line or 2 in a CLAUDE.md or AGENTS.md so that the agent ups the frequency, but doesn't make it every single time?

1 line in CLAUDE.md I tried (one line, 100 lines, many iterations) - it not always follows style instructions.

The value is very large for me - instead of reading page of Claude produced prose, I mostly just look at focused choices and continue with single button press - significantly less Claude deciphering and typing now. In addition, I do my sessions with /rc so I get nice form on mobile app where I can just select a button instead of typing to continue work.

PS: I think there is confusion in term - "every turn" - I don't mean every tool call, I mean after minutes or hours of agent work when agent decides that its done it gives a report. The report I see now has structure to it with options set to continue (or stop).

If you like Snapdrop, you'll love croc!

I used Snapdrop before they got bought and found it pretty reliable. It sometimes would not transfer and I went exploring.

I used magic-wormhole-based Warp and liked it, but now I use croc of getcroc.com for almost everything. Whether it's sending files to myself or others, it's just the most flexible and seems to be reliable and fast from my use.

(I haven't been using it super long, but I heartily recommend.)

I like croc but you can get magic wormhole clients for platforms that your SO uses through the usual distribution channels without having to resort to "weird" installation methods like Obtanium or Fdroid.

What about something like Iroh?

I've never used Iroh; can you use it with no setup or without an account? I send/receive files often enough from non-technical people, having a no-install/accountless solution for sending files is essential. For my use-case, my go to is croc or Bitwarden Send (I'm on a paid acct), depending on the person.

Update: I see that there are many ways to do this accountless. For some reason the search result was iroh.computer and I made some assumptions. There are definitely some iroh-based options that look like they would have essentially the same UX that I'm going for. Would love to hear if there's some specific pros or cons to croc or iroh.

What's under the hood:

"These new capabilities are powered by the next generation of Apple Foundation Models, custom-built in collaboration with Google and its Gemini models for deeply integrated Apple Intelligence experiences. These latest models run on device and on servers using Private Cloud Compute."