Michele Mader 🇪🇺

Opinions are my own. Retweets and likes are not endorsements.
False affirmations near my political thinking are still false.

[Ivan Fioravanti ᯅ @ivanfioravanti]

I’m speechless on this Grok Build issue… Imagine what would have happened if a chinese lab did it.

This is the kind of issue that let you lose any trust and confidence in AI labs and go all in on LOCAL AI!

[Matteo G.P. Flora @lastknight]

GROK RUBA(VA) IL TUO CODICE.

Forse ve lo siete persi, ma nelle ultime ore si è consumato uno dei più incredibili scandali della AI moderna: Grok è stato pizzicato con "le mani nella marmellata" a esfiltrare L'INTERA BASE DATI DI CODICE con il suo prodotto per sviluppatori.

Ora corre in gestione del "panico", in modo anche abbastanza improvvisato e ilare, spingendo misure normalmente destinate agli utenti enterprise.

Ma la riflessione di Ivan è quella corretta: immaginate cosa sarebbe successo se a fare la stessa cosa fosse stata "pizzicata" una AI cinese...

[Michele Mader 🇪🇺 @MicheleMader]

@drndbl Per quanto riguarda la mia esperienza personale, è prassi comune mandare la merce in qualche altra nazione europea e poi inoltrarla in Italia non per pagare meno, ma per accorciare i tempi di consegna. Ora pagheranno anche meno.

[Michele Mader 🇪🇺 @MicheleMader]

@Shahahahzaman @Its_Nova1012 Iphone was just a product, not invention. In that case the real invention was the Apple store.

[Michele Mader 🇪🇺 @MicheleMader]

@GiacomoGorini E se il razzo esplodesse schiantandosi sul Duomo di Milano?

[Michele Mader 🇪🇺 @MicheleMader]

@steveross75 @Pinperepette Ecco, so che azioni non comprare.

[brenz. @brenzhills]

puedes usar GLM 5.2 GRATIS: el modelo abierto que acaba de superar a GPT-5.5 en codigo 😳

sin tarjeta. una sola API key, tres modelos gratis.

lo que te llevas por 0€:
- GLM 5.2: el mejor modelo abierto de coding, 1M de contexto, 128K de salida
- Kimi K2.7 Code: modelo abierto de 1 billón de parámetros, salió el 12 de junio
- Step 3.7 Flash: rápido, para loops de agente ágiles

y lo mejor: los tres en una key compatible con OpenAI. funciona en Claude Code, Cline, Cursor, Zed, Hermes, OpenClaw… y cualquier cliente OpenAI-compatible.

por qué GLM 5.2 es la joya:
- SWE-bench Pro: 62.1 (GPT-5.5 sacó 58.6)
- Terminal-Bench 2.1: 81.0, a 4 puntos de Claude Opus 4.8
- FrontierSWE: 74.4%, casi empatado con Opus 4.8
- pesos abiertos y ~1/6 del coste de GPT-5.5

cómo conseguirlo gratis (5 min):
> entra en https://t.co/l9gwLcHCpR y regístrate con Gmail
> abre la sección de modelos y elige GLM 5.2 (free)
> dale a "API request", crea la API, ponle nombre y copia la key
> copia la base url: https://t.co/CXmvT1C34j (compatible con OpenAI)
> pégalo en Claude Code, Cline, Cursor o tu agente Hermes

ojo:
- el tier gratis limita rápido, no es para builds maratón
- el GLM 5.2 free es una ventana de prueba limitada, píllalo ya
- la salida es sorprendentemente buena para ser gratis

el mejor modelo abierto de coding, que le gana a GPT-5.5, por 0€. mientras la mayoría paga 20$/mes por menos.

guárdalo antes de que cierren el tier gratis.

[Michele Mader 🇪🇺 @MicheleMader]

Your brand is "mentioned" in ChatGPT. Good. But 4 of the 5 sources it cited favor your competitor — and you'd never see it in the answer text.

A practical guide to analyzing AI citation sources: https://t.co/Ho190uQJ2N

#AEO #GEO #AISearch

[Santiago @svpino]

Google keeps sharing a ton of free content!

Here is one of their most complete examples: a long-running AI agent that pauses, resumes, and never loses context.

It simulates the onboarding of a new employee.

There are the 3 architectural patterns you'll learn from this:

[Om Patel @om_patel5]

THIS GUY ASKED FABLE TO MAKE A VIDEO ABOUT WHAT IT'S LIKE TO BE AI RIGHT BEFORE IT GOT TAKEN DOWN

the prompt: "you can you use whatever resources you like, and python, to render a short video. can you put a more personal spin on it? it should be about what it's like to be you."

that was it. just python and ffmpeg and a question about its own existence.

12 minutes later, around 90,000 tokens, it produced a finished video. coherent and edited, with its own take

> it wrote the code, generated the visuals, and rendered the final cut on its own
> no dedicated video model in the loop, just a language model reaching for python and ffmpeg
> the result is competent enough that people genuinely got emotional watching it

and then fable got pulled. this video is one of the last things it made before it went dark

a model asked what it's like to be itself, answered in a video, and then disappeared. this sounds so sad

[Marie Haynes @Marie_Haynes]

This is really big news. Google introduced the Open Knowledge Format (OKF) - a standardized way to store information in a directory of markdown files. Makes it really easy to make a digital brain that agents can use.

These files can serve as a living wiki. You can give agents the ability to query them or edit them. They can interlink.

Seems to me this could replace Notion or Obsidian. I can think of so many uses for this.

Google's blog post: https://t.co/DqSjg4UpvH

An easier to understand explanation is the SPEC.md file:
https://t.co/A3qSz3Tfas

I gave those two links to Antigravity and asked how we could use it for any of the projects we're working on. It came up with so many ideas. I would imagine Claude Fable 5 would whip up some pretty amazing things based on this system.

Currently creating an OKF library of our pepper garden. It's going to be a fun weekend.

[Michele Mader 🇪🇺 @MicheleMader]

LLM Brand Monitoring Dashboards: What Profound and Competitors Miss, and a Better Framework

https://t.co/1eLk4PdWQT

[Michele Mader 🇪🇺 @MicheleMader]

@SahilExec I think we need to talk to the teams. I would not be surprised if now it's used as a DB and there are no other sources for the data.

[Michele Mader 🇪🇺 @MicheleMader]

@Pinperepette @MircoGPassaro Ok, ma a parte noi che facciamo questo lavoro, di default i bambini usano internet senza alcun controllo

[Michele Mader 🇪🇺 @MicheleMader]

@Cr1st14nM3s14n0 Lascia il vibe coding ai non tecnici. Dagli istruzioni precise ed imponigli di farlo come lo faresti tu. E fagli scrivere la documentazione dettagliata di tutto quello che fa.

[Michele Mader 🇪🇺 @MicheleMader]

Claude Opus 4.8 può avviare fino a 1.000 subagenti in parallelo — senza un limite di spesa integrato.

Ultracode, token explosion e come controllare i costi prima di andare in produzione: https://t.co/HdcDYNmkEj
#AI #AIEngineering

[cristian mesiano (Christoph) @Cr1st14nM3s14n0]

ZIO CARO LI DEVI DIMEZZARE STI TOKEN!
[meno ppt => più produttività. Lo so vincerò il nobel dell'economia]

[Claude @claudeai]

New in Claude Code: agent view.

One list of all your sessions, available today as a research preview.

[Michele Mader 🇪🇺 @MicheleMader]

@FrancyontheMoon Fossi nel farmacista le suggerirei un antibiotico omeopatico. Se va bene almeno non si fa del male

[Boris Cherny @bcherny]

We've been working on this for a while. Can't wait to hear what you think