
🇪🇸🇵🇹 Partnering with Startups at @Cloudflare | Phd | Opiniones propias
Spain
Joined January 2023
- Tweets80
- Following1.2K
- Followers155
- Likes139
Carlos Luengo retweeted
so cool hearing about how and why @cloudflare powers @openai chatgpt sites
all the primitives for building a platform in one place: workers for platforms, d1, custom domains… everything you need!
Carlos Luengo retweeted
The Cloudflare's writeup on our php2shell response is up.
Rules went live at 03:02 UTC, about 9 hours before disclosure. Unfortunately the vulnerability leaked early, so we were already watching attackers mutate payloads against the rules and updating as variants appeared.
New rules keep landing in the WAF changelog.
Expect updates through out the day. We will have a more thorough post of what we had to build to better protect people coming shortly.
PHP runs almost 70% of websites. The flaw is in the runtime, so updating plugins won't fix it. Patch PHP.
blog.cloudflare.com/php-vuln…
Hey @evanyou, we need to talk! We'd love to see you at Cloudflare Immerse Madrid 2027 and with @exp8fellowship for a fireside chat alongside @GuliMoreno. Also, huge congrats on your talk at Cloudflare Connect!
Listening to @evanyou at #cloudflareconnect makes it clear that we have entered a much better era for building and deploying code. We will look back on the last decade and wonder how we did it.
Wow @Lovable en nuestro Cloudflare Connect! Que gran momento para el ecosistema startup! Cloudflare el mejor sitio para construir el internet agentico!!!! 🚀
Seeing is believing. Love seeing @FabianHedin live demo what @Lovable built using Cloudflare Workers and AI services. The future is happening now.
Carlos Luengo retweeted
Seeing is believing. Love seeing @FabianHedin live demo what @Lovable built using Cloudflare Workers and AI services. The future is happening now.
Carlos Luengo retweeted
kitesurf
Kitesurf is a browser that runs inside a @Cloudflare Worker. No Chromium, no container. @mattzcarey demoed an agent driving it live at init().
Carlos Luengo retweeted
Tomorrow. Come for the (temporary) coffee, stay for the (durable) objects.
luma.com/k72rpi3a
Carlos Luengo retweeted
Cloud Coffee: "Durable Objects"
Happening now at SOHN, San Francisco. Come say hi!
😬
The new @cloudflare @vite_js plugin v2 drops Wrangler, infers a worker's env, and exports types from a TypeScript config.
James Opstad tells us more about that and the new cf CLI, where cf dev and cf build delegate straight to Vite 🔥
@ViteConf 2026, Thu 15 Oct, 14:00 UTC. Free and online.
viteconf.org/
Carlos Luengo retweeted
¿Qué modelo tengo que usar? O, ¿qué modelo abiertoes similar a Opus? Nos preguntáis esto muchísimo, así que me he puesto a escribir una especie de guía sobre :
"Qué modelos abiertos usar en lugar de Claude, GPT o Gemini"
No hay equivalencias 1:1, pero sí que hay una tabla de los modelos de Claude, GPT y Gemini y su homólogo abierto más cercano.
Son modelos de propósito general, pero merece la pena pensar primero en la carga de trabajo a resolver antes que en el modelo. La carga de trabajo te debería llevar al modelo más óptimo para resolverla.
Por ejemplo, para coding, seguridad, infra... GLM 5.3 nos funciona genial
Pero para implementar un plan de coding ya cerrada, tiramos de modelos más rápidos y "baratos" como puede ser ds4.1 flash
Hay un poco de explicación de cada carga de trabajo y su mejor modelo abierto en el artículo. Además de la estrategia de routing que utilizamos.
Nuestro setup ahora mismo es GLM 5.3 + Deepseek v4.1 flash y va volando. Yo para tareas creativas tiro de Astra y Opus a veces.
Tenéis el artículo completo aquí: helmcode.com/es/posts/altern…
Decidme si preferís leerlo en la web o como artículo de X, que quiero ver qué tiene mejor distribución.
Espero que os sirva!
Carlos Luengo retweeted
Por primera vez en mucho tiempo, Saferlayer debería funcionaros sin VPN aunque haya fútbol. Y seguimos con Cloudflare 💪
saferlayer.com
Ayer en el meetup de OpenAI de Madrid. Más rápido que la vuelta de Sam Altman: Desplegando con OpenAI en Cloudflare. Usar codex para desplegar tan solo haciendo @cloudflare
Carlos Luengo retweeted
We're live on Product Hunt today.
Zernio started as a social API -> One integration instead of sixteen OAuth flows. This is still live.
Suddenly, the product got bigger.
Same API now covers ads, WhatsApp, messaging and calls. Agents can publish, reply, text and pick up the phone without you wiring each channel.
This is what launch today is about:
Marketing infrastructure for products and agents.
If you've ever started a social integration and watched it turn into a six-month maintenance job, this is the alternative.
Upvote or drop a comment. We'll be in the thread all day.
producthunt.com/products/zer…
Carlos Luengo retweeted
Today we're repositioning Zernio.
Agents won't just build the product. They'll run how it reaches customers.
That's not a theory but it's already happening.
We started with social integrations.
Now it's ads, WhatsApp, messaging, and calls. Same API.
So we're calling it what it is: marketing infrastructure for products and agents.
































