All marketers ~~are liars~~ tell stories.

What metrics are you tracking specifically?

Opposite problem here - all building, no marketing. Now struggling with GTM because nobody knows we exist.

The real trap is thinking it's either/or. You need both engines running.

As others mentioned, you can't stop them from doing this. It's a cutthroat game where companies will try to poach customers however they can.

Best defense is making your customers so happy they won't switch, but tactically:

- Create a private Slack/Discord for paying customers only (harder for them to infiltrate) - Move community to a platform with anonymous usernames (Something similar to Reddit-style forums or Discord with display names) - Add some soft vendor lock-in -- a bit evil but sometimes necessary - just maintain the balance between retention and not being user-hostile

On a positive side, if a VC funded startup is this desperate to poach your users, you're clearly doing something right.

- from fellow wannabe bootstrapper

IDM - Internet Download Manager -- Only available for Windows, never saw a proper alternative for Linux/MacOS with the same reliability and browser integration.

Though no longer download videos/files enough to complain about it.

I end up using yt-dlp on macOS/Linux to download videos if needed. It’s worked on pretty much everything I’ve tried it on so far. Though admittedly not at seamless as a button being injected into the page itself.

Still on youtube-dl here - didn't know yt-dlp was the actively maintained fork.

Does youtube-dl still work? There seems to be a cat and mouse game with these things. I won't update it for a while and it will break, but it always works again after I update. I figured without active maintenance youtube-dl would be pretty useless by now.

Couldn't relate more.

I built this Chrome extension using similar logic to youtube-dl: https://chromewebstore.google.com/detail/audio-only-youtube/...

Eventually couldn't keep up with maintenance and leave it, sigh.

Curious how you're thinking about the MLE role evolving with LLMs disrupting so much of traditional ML work?

BTW, try Claude Code for your next project - you will love it! :)

In my experience, LLMs tend to take noticeably longer to process images than text.

It has to get the image data first, basically just IO time before processing it

IIRC there's pre-processing (embedding/tokenization?) before feeding images to LLMs?

Hit this issue optimizing LLM request times. Ending up lowering image resolution. Lost some accuracy but could bear that.

I wonder if these stay in the prefix cache?