pea

I just spent 10m trying to use this.

It is the most awful product experience I’ve had this year. How could they fuck it up this badly?

I actually think Amazon is very well placed:

- Single point of billing and vendor is still king in enterprise - look at how much of Ant's revenue flows through AWS Marketplace

- With OSS, models may move to a more of an AWS-shaped problem: host with 5 9s, provide the infra and shovels, do it as cheap as possible, fit them neatly into rest of AWS stack. I think this is actually a huge risk to OpenAI and Anthropic, as more commodization gives hyperscalers more advantage

- Still owns a huge chunk of Anthropic and leverage over compute

With this in mind, I'm not sure why they need a frontier model vs. the playbook they did with Linux (i.e. Kimi-K3-AWS-Enterprise-Edition). They are kind of playing both sides right now I guess, but could really stir things up if they wanted to.

I think we'll see it more and more... open-ended Q&A / novel work is everyone's first experience of LLMs, but I don't think it's actually representative of most of the work people do. Coding kind of sits in the middle as it's novel but semi-structured.

Chat is great for novel open-ended work, but is terrible at concurrency - i.e. managing > 1 agent runs. As agents get cheaper, I suspect we'll see more more bounded processes where you want to run lots of items through them at once, and thus Jira-like things... (we ended up having this problem and building something like this, albeit more medieval themed [0])

We also found the other piece that Jira-style solves (and that chat makes super hard) is multi-player. I've yet to see a good implementation of a multi-player chat solution to agents working concurrently; and we found that an async model is the only sane solution to this.

[0] https://pumpup.com

I had to build a diffing engine for PowerPoint files in my first start-up. I initially tried to do this at the OOXML level, and one the weirdest things I found about the format is that the GUIDs are mutated by PowerPoint even if there have been no changes. The serialized OOXML is not really canonical and constantly changes as clients open and save it.

The simplest solution ended up being rendering images of the slides (not easy) and doing it with a slider.

That's a great idea with the e-reader.

I would love for someone to design a great case with a built in e-reader on the back. Most of the existing ones seem geared for displaying colour photos; I'd like one built for reading, powered by NFC. It should easily sync up with the phone itself to send over articles, books, etc. Bonus if I can write on it.

I'm exactly the same. This one really give me the ick: https://www.youtube.com/watch?v=usYC_Z36rHw

I was wondering the same for vega-lite, which is relatively high level, declarative, and looks similar to their syntax.

The shift may come when your blessed cloud platform provides competitive enterprise-ready hosting of the oss models

Looks cool but your home page heavily lags on my mbp m3 pro - you should maybe be lazy loading vs loading all your components upfront

I agree with mostly everything you’re saying; but it’s not uncommon to be processed via your local consulate, even if you are already living in the US.

This is usually just for the final issuing of the GC, and where USCIS approval has already happened (for instance, on an EB1A).

People frequently do this so they don’t have the travel restriction. Source: I just did it.

Good fun! Spacebar should launch, r should reset, maybe I'm imagining it, but it's a lot slower and thus easier on my phone than my MBP.

Oracle hire some extremely good people and pay very well

The darkest UX pattern I have ever hit is trying to cancel Google Workspace; whereby they disable the scrollbar on the page so you cannot actually get to the cancel button.

Decapod

What sort of things do you notice?

Thank you!

maybe productive isn’t the right word - more like satisfying and constructive. Right now I might play blitz chess or end up checking the news or X. But when I’m in the middle of a good book or some long form content, I’ll use that time to read and it’s much more fulfilling.

And I think breathing meditation is an excellent idea! I didn’t only mean phone-specific activities at all! Thank you

I think it makes sense given the following:

- Your brain has been trained extensively to recognize faces / people. Even very small babies can do this.

- Your brain processes a large amount of mostly noise, and sometimes mislabels noise as objects, which trends towards face-like things (see: seeing faces in clouds, people in shadows etc.) Various classes of substances make this effect more noticeable (even stimulants, including caffeine)

- The jump from that to 'elves' is largely just cultures have some form of small magical person.

This is really cool, I enjoyed going through them in this form. Thanks

Didn't they also just shut down Vertex and Gemini APIs to launch a new unified API this week?

It’s a shame because people forget how good IBM research was back in the day. I do wonder if they still have great people in those r&d labs, or if they all left.

We had one of these today in Sausalito!

The way Gemini (which is incredible) has been haphazardly shoved into gsuite products seems devoid of any kind of common sense. Even very basic obvious things via Gemini in Gmail fail. It’s ironic because it’s the one place I’d love a decent model.

I got a nasty one of these recently. Attacker had my wife’s CC, and made purchases they knew would flag our bank and look sus and pop up on the app. Then I get a call from my “bank”, correct number, I ask them to verify and they say look at the number. All they ask for is the customer support code in my banking app to cancel to fraudulent transactions. At that moment I insist on calling back and they hang up, but I was very close.

I wonder if you could spend a few million on promoting adblockers to justify a short position on Google or Meta.

I feel like the opposite is also true. People view Java as enterprise Java from 2008; a clunky outdated language for bad, verbose spaghetti code.

In fact, a lot of the most interesting plt and compiler r&d going into real world applications is on the jvm (project loom, graal etc), and the features of modern Java (pattern matching, records, etc) make it a great choice for lots of projects that aren’t legacy enterprise apps.

For something slightly more expensive check out NTH (now defunct micro brand but easy to find on eBay) or Squale.

For vintage (ignore their new stuff), Enicar, Eterna, Roamer, Rodania, Cortebert, Wakmann, Mido, Certina, early Girard Perregaux, etc. all can be had cheap. Bunch of excellent brands wiped out in quartz crisis that can be had for next to nothing.

Thanks. I don’t have mullvad either - anyone can point their dns at their Adblock dns without paying for their VPN