Umberto Raimondi

Making computers do things mostly under duress |

[Umberto Raimondi @uraimo]

@redtachyon I'm starting to think that it's a concept too hard to understand for somebody living in a world of fintech crap.

[Umberto Raimondi @uraimo]

@tekbog … to the permanent underclass?…

[Umberto Raimondi @uraimo]

Why not "Slop-o-matic"

[Negligible Capital @negligible_cap]

*META DEBUTING NEW APP WITH FEED OF VIBE-CODED GAMES: INSIDER

$META is rolling out a new app called Pocket, which will allow people to vibe code video games

[Umberto Raimondi @uraimo]

After Mazda, some new cars from Renault will come with Gemini as default assistant, finally something usable. Maybe it’s too soon but it could be nice to have a standardized “MCP” with different access levels to support any external assistant/agent.
An Odb for the agentic era.

[Umberto Raimondi @uraimo]

@camsoft2000 Black hole I suppose, but I too dream of a better world

[Umberto Raimondi @uraimo]

@vimtor Sigh

[Umberto Raimondi @uraimo]

@arkuy99 I will now start using that X translation “may tens of millions bring you safety and joy” as a greeting with my friends 🥹

[Wenting @zephray_wenting]

Might sound a bit cursed, but I turned the M5Stack PaperS3 into a GameBoy emulator:

[Umberto Raimondi @uraimo]

The stream deck has a stable python library to drive the device, show images and text on all the screens and read all the events included status of the dials and touch screen.

[Umberto Raimondi @uraimo]

People of Pi, just so you know if that is what I think it is, you can already build yourself a pi-deck as I did months ago.
It's nice to have something physical when the terminal is not in focus but it gets boring fast.

[OpenAI Developers @OpenAIDevs]

Your favorite Codex shortcuts are getting an upgrade.

July 15th.

[Umberto Raimondi @uraimo]

The best feature here is having buttons that the agent can configure depending on the context, and having a small screen where for example requests for input or notifications can appear.

[Umberto Raimondi @uraimo]

@ivanfioravanti This or a Fiat Pandina... hard choice.

[Umberto Raimondi @uraimo]

Be like goose

[Ed Zitron @edzitron]

SoftBank’s investor presentation is one of the greatest things ever made. I’ve been thinking about it all day. These are the real slides shown in a speech where Masayoshi Son said he wouldn’t retire for at least another decade. The goose stuff is perfect.

https://t.co/sk9cDhdWIE

[Ugur Toprakdeviren @toprakrun]

MiniSwift Studio is now an installable, fully-offline Swift IDE.

The question that started it: why do you need a Mac and Xcode just to try SwiftUI?

MiniSwift Studio runs a Swift compiler and runtime entirely in your browser — WebAssembly, 100% client-side, no server round-trips. Live SwiftUI preview, a real breakpoint debugger, SwiftData, Foundation, Metal, GameKit etc.

And as of today you can install it like a native app and use it completely offline.

The part I'm proud of: the live debugger relies on SharedArrayBuffer, which needs cross-origin isolation.

Making that survive offline, served from a service-worker cache instead of the network, was the hard part.

Now you can board a plane, launch it from your dock, and compile + run + step through real SwiftUI with zero connection.

Also new:

• Open a real folder from disk and save back to it (File System Access)

• ~16 MB cached once → instant, and offline forever

• A clean update flow that never drops your unsaved code

No Mac. No Xcode. No server. Just a browser.

Try it: https://t.co/6jRu741qZG

Install it: hit "Install app" on https://t.co/CHKxAzOcuB

Would love your feedback.

#Swift #SwiftUI #WebAssembly #PWA #iOSDev #WebDev #DeveloperTools

[François Chollet @fchollet]

AI in 2040 will not be built on the stack we are using today. It will be much closer to optimal. The current stack has 3-4 orders of magnitude of data inefficiency and 4-5 orders of magnitude of compute inefficiency.

Near-optimal AI is what symbolic learning will deliver.

[Umberto Raimondi @uraimo]

Those who don’t get this instantly are part of the problem.

[Mitchell Hashimoto @mitchellh]

Got em. I poison my AGENTS.md (and other things like code comments) all over the place with prompt injections like this to find people who don't review their code and sling it off to another human. Catches folks all the time and then its an instant ban.

As I've said, I don't care if you don't review your own code. But if you're submitting code to an OSS project and crossing a human boundary, it is simple courtesy to do some human review.

[Jeremy Howard @jeremyphoward]

Wow.

@Zai_org GLM 5.2 is a marvel! It is *at least* as good as Opus 4.8 and GPT 5.5. It's super fast, inexpensive, and not too verbose.

It responds with nuance and judgement, & handles long context VERY well.

I've never experienced an open weights model like this before.

[Arthur Mensch @arthurmensch]

First, we have a nice model coming this summer – we hope it will delight and surprise in a few capabilities. This will be the start of a new family of models, fat indeed, but sparse. We're opening up an early access program in July for key partners in research, government and the industry.

[Gökdeniz Gülmez @ActuallyIsaak]

MLX-LoRA-Studio a MacOS native, user friendly and customisable as hell app for training LLMs locally on Apple silicon. full release will be coming. GH Repo: https://t.co/SLXaavoOwA

The website is online as well: https://t.co/KPjreJ2mBF

[Arthur Mensch @arthurmensch]

It's actually le gros chaton