Simon Karan

@ChShersh operator overloading

gun to your head your favourite C++ feature

@johncrickett To achieve what Jensen Huang is suggesting, a human in the loop is still needed to orchestrate the LLMs at some level

“I think we’ve achieved AGI.” --Jensen Huang Where AGI is: "start, grow, and run a successful technology company that’s worth- worth more than a billion, more than a billion dollars." But: “A lot of people use it for a couple of months and it kind of dies away. Now, the odds of 100,000 of those agents building Nvidia is zero percent.” So which is it? My take: LLMs (and all of AI) are great tools, we're far from AGI.

Kudos to @ouraring for their amazing customer support agent. It was able to issue a replacement ring by automatically checking my ring’s status and confirming my delivery address in under 5 min 😱 A true masterclass on automated support 👏

@DFB_Team Sané 💤💤💤💤💤💤💤

Unser Kader für die anstehenden Länderspiele gegen die Schweiz und Ghana 🤝 _____ 🇨🇭🇩🇪 27. März - 20:45 Uhr - RTL 🇩🇪🇬🇭 31. März - 20:45 Uhr - ARD #dfbteam

Imagine a world were Siri is powered by OpenClaw @steipete

@johncrickett The basic software engineering concepts are more important than ever.

I've been documenting my views on AI-assisted coding. I believe in strong opinions, weakly held, so I'd like thoughts, feedback, and challenges to the following opinions. On AI and AI-Assisted Coding: → The agent harness largely doesn't matter. The process should work with all of them. → Most AI-assisted coding processes are too complex. They clutter the context window with unnecessary MCP tools, skills, subagents, or content from the AGENTS file. → A small, tightly defined, and focused context window produces the best results. → LLMs do not reason, they do not think, they are not intelligent. They're simple text prediction engines. Treat them that way. → LLMs are non-deterministic. That doesn't matter as long as the process provides deterministic feedback: compiler warnings as errors, linting, testing, and verifiable acceptance criteria. → Don't get attached to the code. Be prepared to revert changes and retry with refinements to the context. → Fast feedback helps. Provide a way for an LLM to get feedback on its work. For example, tests, compilers and linters. → Coding standards and conventions remain useful. LLMs have been trained on code that follows common ones and to copy examples in their context. When your code align with those patterns, you get better results. On Software Development: → Work on small defined tasks. → Work with small batch sizes. → Do the simplest possible thing that meets the requirements. → Use TDD. → Make small atomic commits. → Work iteratively. → Refactor when needed. → Integrate continuously. → Trust, but verify. → Leverage tools. → Don't get attached to the code. What are your strong opinions on AI-assisted coding?

@CodelyTV Igual el ejecutar una CLI para un agente génerico se tendría que hacer a través de MCP. Por lo tanto no necesariamente es lo uno o lo otro. Más bien ambos se complementan

El CTO de Perplexity acaba de decir que van a dejar atrás los MCPs y en su lugar se van a centrar en APIs y CLIS. En respuesta, más empresas han empezado a posicionarse de manera similar. Así que después del boom que hubo el año pasado… ¿están dejando de tener sentido los MCPs? Respuesta corta: no. Respuesta larga publicada en nuestro blog cdly.to/mcp-muerto-t

@DougieCritchley It‘s disrespectful to Luis Diaz to have his name to closely written to Sané‘s.

Fair play Bayern Munich. A lot of people, myself included... questioned whether spending €75m for a then 28 year old Luis Diaz was a sensible, long-term investment. But how Bayern have benefitted from signing an experienced, win-now player like Diaz. He's just so many levels above Coman and Sane, and his arrival has even sparked a revival in form in Gnabry. Olise was the real star last night, but Diaz completed 8 dribbles, created 2 chances, had 4 shots and provided an assist. That's his 37th goal or assist now in 36 games... Been brilliant.

@iMiaSanMia Bayern needs to buy a replacement for him. Never really got to recover from his injuries

Alphonso Davies is injured again, off in tears. Bischof comes on

@vchennai2 If you are into the classic Jetbrains IDE experience the claude code plugin works great. They also recently added support for cursor

I've switched back to cursor from Claude Code Claude code makes it far too easy to lose codebase context since you're only looking at diffs For an application that requires precision that starts becoming a problem real fast Curious if anyone found a good way around this

@oarangov @villamzr En tu experiencia que agente de desarrollo sería el mejor para Java? Entiendo lo que te dices con „IDE-driven“ y por eso creo que las recientes integraciones de Jetbrains con Cursor, Claude Code y Codex pueden ser interesantes

@simonkaran13 @villamzr No es malo, sólo que no contempla todas las capacidades que giran en torno a Java: Manejo de classpath, resolución de símbolos, maven/gradle, profiler, refactors, entre otros. Se siente corto. Java es muy "IDE-driven", no es sólo archivos en un directorio como podría ser JS/TS.

Justo ayer conocí una empresa que usan disque Github Copilot. El más manco de las IAs. (?)

@oarangov @villamzr Que tiene de malo cursor para proyectos Java?

@villamzr Conozco empresas que quieren ser AI First sin siquiera entender qué implica o requiere esto. Otra que paga Cursor para proyectos Java, y así. En Latam hay de todo.

@Solfamido_81 @jesus_osorior 15 champions y 2.5 penalties

@jesus_osorior ¿Eso en Bernabéu cuántos son?

¡El tamaño real del asteroide que extinguió a los dinosaurios!

@MoureDev La integración con Claude Code me va muy bien la verdad.

Justo estaba hablando con un compañero de que nos daba pena que una empresa como JetBrains, que ha creado herramientas tan buenas durante años, se hubiera quedado atrás en la guerra de los IDE y la IA. Con ganas de probarlo.

@jamonholmgren MCP are good for their plug and play nature. So what would be the best of both worlds? An MCP that exposes a CLI 🤔?

The vast majority of MCP servers can just be replaced by CLIs. Change my mind.

@TouchlineX Bernal clears him anyway

🚨💣 𝐄𝐗𝐂𝐋𝐔𝐒𝐈𝐕𝐄: Frenkie de Jong has agreed with the Dutch National Team that he will not take a single risk for Barça in the final stages of the season. His main priority is the World Cup after missing out on the EURO'S. He will refuse every painkiller and injection Barcelona will offer him, he will ONLY play for them if he feels 100% fit, not 99%.

@TechByTaraa Slow Wi-Fi

I'm a Vibe Coder, scare me with a word.

@awsdevelopers AWS

Reply to this tweet with "AWS" and we’ll tell you which AWS Service you are

@ChekosWH Al final las hackathons la gana el que mejor vende su idea. No el que tenga el mejor proyecto técnicamente

hoy anunciaron los ganadores del hackathon de Anthropic en el que participé y nos enseñaron sus demos que se suponía eran videos de hasta 3 minutos explicando que construiste y que hace estos vatos parecían anuncios del superbowl no mamen yo todo wey grabandome solo en zoom jajaj

@fanaticodeport @giraltpablo De hecho no está tan frio, para que haya nieve no puede hacer tanto frio. Ahora está entre 0 y 2 grados. El frio real como dijo James es de -10 grados o más

@giraltpablo Aplacen ese partido, con razón James Rodríguez se fue por el frío

Impresionantes imágenes del entretenimiento de Bayern Múnich TOTALMENTE CUBIERTO DE NIEVE