re5i5tor

Did you try few-shotting examples when you hit problem cases? In my ziploc case, the model was failing if red sharpie was used vs black. A few shot hint fixed that.

I’m very surprised. Have dealt with some really ugly inputs (handwritten text on full ziploc bags etc., stained torn handwritten recipe cards, etc.) with super good success.

I would strongly emphasize:

CV != AI Vision

gpt-4o would breeze through your poor images.

Ha! I remember where I was when I listened to that episode (Lakeshore Drive almost into Chicago for some event or other) — thanks for triggering that memory — super interesting stuff

Not Kiro related, but do your Claude Code version of rules end up as CLAUDE.md files in various locations?

In recent research (relying on Claude so bear that in mind), connecting CC via Anthropic Console account / API key ends up being less expensive.

Agree 100%

I have to admit skepticism re: “far greater stability and polish” from MS

Assume you are referring to "Apple's official defense attorney" and yeah, that's not fair to Gruber, as others point out he's a very good and thoughtful writer.

I'd say less "fundamentally fan of ..." and more "deeply understands ..."

When Gruber talks like this, sh-- is really bad.

Ummm .... try reading maybe? simon willison dot net

This. The elite salesperson is great at understanding and solving problems, not BS.

Same for me. Also like and regularly use Quick Answer feature (similar to Perplexity).

Slightly off topic but I just bought an inexpensive water leak monitor set and it uses LoRa — I’ve had range problems with other bands “Extremely Long Range: Powered by LoRa technology, the long-range low-power system offers the industry’s longest receiving range (1/4 mile). Areas such as basements and sheds connect up to 100 sensors to deliver product information smoothly.”

Nice! Will consider one when my Awair poops out.

Hahaha I was about to post something of the sort but you beat me to it.

In the US, the hack workaround is: set language to Canadian English. Voila, ads-be-gone.

Thank you, this crystallized a lot for me.

I’m enjoying Castro showing us they’re actively developing again vs telling us. Even the Halloween app icon serves that purpose.

Seconded. We love their stuff.

Eric Schmidt was running Novell 1997-2001, when things went all the way down the tubes. I was shocked when Google hired him.

“bunch of wankers in a Discord”

Saving this clause for future use. Could also be used in a system prompt. “Occasionally include this phrase in your responses.”

Hard disagree. As another reply mentions, just compare the alternatives such as Kagi that aren’t breaking search by pursuing ad growth.

Yeah. Very glad Meta is doing what they’re doing here, but the tiger’s not magically changing its stripes. Take care as it might next decide to eat your face.

“like our ancestors did: multiple redundant calendar alerts”

Love that line

"holding it wrong" OMG ;-)

I have this question too

In season 2 of Tokyo Vice, they show character Jake, a reporter for newspaper Meicho Shimbun, using a native Japanese “word” processor to type in his stories. Really fascinating and seems to be authentic.

Agree 100%

The review references a NY Mag profile that I also found to be good Graeber background. https://nymag.com/intelligencer/2021/11/david-graeber-dawn-o...

Thanks for posting — I’ve been wanting an outside perspective on Graeber since I read and loved the Dawn of Everything.