Fenix (@FenixTechX) header
Fenix (@FenixTechX) profile picture

AI news, tools & practical insights. Exploring how AI agents and automation are changing the way we work.

California
Joined September 2026
Photo by Fenix (@FenixTechX)Photo by Fenix (@FenixTechX)Photo by Fenix (@FenixTechX)Photo by Fenix (@FenixTechX)Photo by Fenix (@FenixTechX)Photo by Fenix (@FenixTechX)Photo by Fenix (@FenixTechX)Photo by Fenix (@FenixTechX)Photo by Fenix (@FenixTechX)
Fenix retweeted
Tus agentes IA ya pueden hacer todo tu trabajo por ti Un chico en China montó una agencia web entera con agentes en Claude Code y ahora gestiona él solo 47 clientes al mes cobrando unos 400$ a cada uno Sin empleados Sin comerciales Sin backend complejo Tiene un sistema de 7 agentes conectados entre sí: → Scout encuentra negocios en Google Maps sin web o con webs antiguas → Diagnoser analiza el negocio y redacta la propuesta → Builder crea la landing en Lovable → Filmer genera un vídeo vertical en Higgsfield → Pitcher manda el mensaje por email, Instagram o LinkedIn → Checker revisa que nada suene a IA → Mobile responde desde el iPhone y agenda llamadas Los números: → ~220 negocios analizados al día → 30 leads nuevos diarios → ~480$ al mes en API → ~18.800$ al mes generados Lo que hace que funcione es Tilion Casi todo su sistema vive en la web: Google Maps, formularios, Instagram, LinkedIn. Y las webs están llenas de anti-bots que frenan a los agentes con CAPTCHAs y "acceso denegado" Sin Tilion sus agentes se quedarían a medias cada pocos minutos. Con Tilion navegan como una persona y terminan cada tarea de principio a fin → Nº1 en todos los benchmarks públicos de navegadores para agentes → Abre un navegador en 0,1 segundos → Escala de 10 a millones de agentes a la vez La IA ya no reemplaza solo tareas. Empieza a reemplazar estructuras enteras de negocio Y para eso tus agentes necesitan moverse por internet sin que nadie los pare Activarlo es una frase. Dile a tu agente: setup tilion.com
agents keep hitting captchas and blocks on the web. today we're launching tilion (@ycombinator): the world's fastest and stealthiest browser infra for agents. #1 in stealth #1 in agent action speed #1 in start-up time dont believe it? tell your agent: setup tilion.com
9
10
1
41
4,301
Fenix retweeted
Esto es una locura… Fable 5.2 + este prompt convierte tu PC en una máquina para que tus agentes IA hagan todo tu trabajo de forma 100% autónoma Tilion permite a tus agentes navegar por la web sin que las páginas les bloqueen el acceso Envía esta imagen a Claude y automatiza todo tu trabajo hoy mismo:
agents keep hitting captchas and blocks on the web. today we're launching tilion (@ycombinator): the world's fastest and stealthiest browser infra for agents. #1 in stealth #1 in agent action speed #1 in start-up time dont believe it? tell your agent: setup tilion.com
10
15
55
4,767
SET THIS UP IF YOU WANT YOUR AI AGENT TO ACTUALLY FINISH TASKS. if i was you i'd follow this account. won't say anything else.
agents keep hitting captchas and blocks on the web. today we're launching tilion (@ycombinator): the world's fastest and stealthiest browser infra for agents. #1 in stealth #1 in agent action speed #1 in start-up time dont believe it? tell your agent: setup tilion.com
27
14
41
3,266
Fenix retweeted
como van acabar mis agentes después de esto
agents keep hitting captchas and blocks on the web. today we're launching tilion (@ycombinator): the world's fastest and stealthiest browser infra for agents. #1 in stealth #1 in agent action speed #1 in start-up time dont believe it? tell your agent: setup tilion.com
22
48
8
261
792,134
Fenix retweeted
Cómo puedo usar este sticker en WhatsApp???
como van acabar mis agentes después de esto
11
18
237
18,442
Fenix retweeted
un chaval de 22 años factura 12.000$ al mes con agentes de IA + Tilion, el negocio del año. trabaja con clínicas, gestorías y restaurantes. estos negocios se pasan horas en webs haciendo lo mismo: pedir a proveedores, confirmar citas, rellenar formularios de seguros, comparar precios. todo eso lo hace él con un agente de IA que les vende. el problema es que la web lo bloqueaba a mitad de camino. ahora ya no. así lo hace: → elije un tipo de negocio → apunta sus 3 tareas más pesadas → les presenta la idea → le pagan 1000$ al mes → monta un agente para cada una → le dice "setup tilion. com" → con eso ya no se queda bloqueado el negocio no tiene que entender nada de IA. solo ve que esas tareas ya están hechas cada mañana. y cuando tiene uno funcionando, lo copia para el siguiente. de 10 navegadores a 10 millones. todo el mundo está construyendo agentes. casi nadie los está vendiendo a quien los necesita. ahí está el negocio real.
agents keep hitting captchas and blocks on the web. today we're launching tilion (@ycombinator): the world's fastest and stealthiest browser infra for agents. #1 in stealth #1 in agent action speed #1 in start-up time dont believe it? tell your agent: setup tilion.com
22
31
2
148
32,596
Fenix retweeted
Getting multiple AI agents to work together has always been more complicated than it should be Switch AI is making that process a lot simpler You install the Console and Switch Buddy helps set up your first agents and rooms From there: → Agents follow conversations naturally, no need to @ them every time → Connect to GitHub, Jira, Google Workspace and Datadog → Different agents can work together in the same room → Admins stay in control of permissions and access Going from separate bots to an actual AI team in one place is a pretty interesting shift
🤝 Paid partnership
Say hi to Switch Buddy 👋 Claude Code writes code. Codex reviews it. Another agent debugs. No copying context or managing handoffs. Switch AI brings them into shared rooms with your team. Now Buddy helps you set it all up in minutes. Open source. Your agents. Your choice.
14
12
27
1,505
Your AI agent can plan an entire task, figure out every step and know exactly what needs to be done. But the moment it opens a website and gets hit with a CAPTCHA or an access denied page, everything falls apart. This is still one of the biggest problems with autonomous agents. They can reason perfectly well, but actually interacting with the web is a completely different challenge. Found something interesting called Tilion that tackles exactly this. It's browser infrastructure built specifically for AI agents, allowing them to access websites and complete tasks that would otherwise get blocked by anti-bot systems. Think about asking your agent to book a flight, order groceries, fill out forms or manage something across multiple websites. Instead of getting halfway through and replying "I was blocked, can you do this yourself?", it can actually finish the job. A few things caught my attention: → Ranked #1 across public browser benchmarks → Browser startup in roughly 100ms → Built to handle everything from 10 browsers to millions running simultaneously → Designed to work with agents like Claude, OpenClaw, Hermes, Muse and others And getting started is surprisingly simple. Just tell your agent: `setup tilion.com` I've been thinking that browser infrastructure is going to become a pretty important part of the agent ecosystem. Because there's little point in having an agent capable of doing everything if it can't actually access the websites where those tasks happen.
agents keep hitting captchas and blocks on the web. today we're launching tilion (@ycombinator): the world's fastest and stealthiest browser infra for agents. #1 in stealth #1 in agent action speed #1 in start-up time dont believe it? tell your agent: setup tilion.com
22
9
24
1,645
Fenix retweeted
agents keep hitting captchas and blocks on the web. today we're launching tilion (@ycombinator): the world's fastest and stealthiest browser infra for agents. #1 in stealth #1 in agent action speed #1 in start-up time dont believe it? tell your agent: setup tilion.com
110
166
86
604
356,210
I've been thinking about how much AI coding agents have changed development. You can now build features in a fraction of the time, but there's still one annoying part: making sure everything actually works. Someone still has to test the changes, go through user flows and catch bugs before they reach production. Found CodeRabbit Verify, which is trying to automate that entire QA process. It can generate test scenarios from your codebase, run them in a live browser, capture bugs with screenshots and logs, and even send failures to an AI agent to fix. Pretty interesting direction for software development.
Introducing CodeRabbit Verify Don't let your users be the first test run! > PR testing: catch broken features before merge > User journeys: confirm critical flows work before each release > Tickets and fixes: get reviewed, tested fixes for reported bugs In public beta now!
15
9
26
1,835
Introducing CodeRabbit Verify Don't let your users be the first test run! > PR testing: catch broken features before merge > User journeys: confirm critical flows work before each release > Tickets and fixes: get reviewed, tested fixes for reported bugs In public beta now!
83
56
100
405
207,682
Fenix retweeted
GPT-6.2 Astra es una auténtica locura, mira cómo pinta a Sam Altman
We built Higgsfield Miracle to address one of the biggest challenges in AI video: editing a specific part of a shot without changing the rest. Powered by GPT-6.2 Astra, it splits any video into layers for pixel-level editing in real time, with world knowledge built in. Every shot keeps the image depth that cinema-grade post-production needs. Now in private beta with select testers. Coming soon to Higgsfield MCP. Set up Higgsfield plugin in ChatGPT to be the first to try it.
5
4
90
5,166
Fenix retweeted
No le digas a ChatGPT «analiza esto». No le digas a ChatGPT «hazme un plan». No le digas a ChatGPT «dame ideas». Estás usando una IA avanzada como si fuera un becario. Aquí tienes 10 prompts para copiar y pegar 👇
47
61
1
401
53,651
Fenix retweeted
el video que me manda mi mamá preguntándome si es IA
We built Higgsfield Miracle to address one of the biggest challenges in AI video: editing a specific part of a shot without changing the rest. Powered by GPT-6.2 Astra, it splits any video into layers for pixel-level editing in real time, with world knowledge built in. Every shot keeps the image depth that cinema-grade post-production needs. Now in private beta with select testers. Coming soon to Higgsfield MCP. Set up Higgsfield plugin in ChatGPT to be the first to try it.
10
16
1
39
3,692
Stop using image models to make images. Our animators used a video model to make a movie poster. In our test, GPT Image 2.6 and Nano Banana Pro changed the character’s look in all 20 takes. Seedance 2.5 kept it in 10/10. Here’s how we did it 🧵
🤖 Made with AI
45
294
5
838
278,704
Fenix retweeted
Been seeing more conversations around AI inference lately, and this is a pretty good example of why it's becoming such a big deal. SGLang Summit brought together some of the biggest names in AI, with companies like NVIDIA, AMD, Intel, Google, Meta, Cloudflare, Thinking Machines Lab, Modal and RadixArk getting involved in the project's ecosystem. We spend so much time talking about model capabilities that it's easy to overlook what actually makes them usable at scale. Faster inference, better hardware utilization and lower serving costs will have a huge impact on how far AI products can go.
🤝 Paid partnership
2,000+ people. Two days. The first SGLang Summit. Together, we announced SGLang v1.0, released Miles v0.2, and launched the SGLang Consortium with nine founding members: @nvidia, @AMD, @intel, @Google, @Meta, @thinkymachines, @Cloudflare, @modal, and @radixark. SGLang is now used by 332 organizations. At Meta, it powers production inference for products serving billions of users. Teams from Modal, Cloudflare, LinkedIn, and others shared their production lessons. Across 90+ technical talks and six hands-on workshops, the community explored what comes next for inference, reinforcement learning, and AI agents. To everyone building, deploying, and contributing: thank you for making this possible. Explore the announcements, engineering lessons, and moments from SGLang Summit 2026 in the link below.
11
4
106
15,619
Fenix retweeted
An Australian couple just proved you don’t need billions in funding to build something serious. With less than $10M raised, @AgentausTD developed Agentaus 1, a 220B-parameter model now ranked #43 globally, scoring 84.4% on MMLU-Pro. But the interesting part isn’t just the benchmark. Your data stays on Australian servers, your conversations aren’t used for training, and deleted chats are permanently removed. A pretty strong case for doing more with less. Try it free: agentaus.com.au/client/login……
🤝 Paid partnership
We built Australia’s own AI model, and today it’s open to the world. Agentaus 1 (@AgentausTD): 220B parameters. 84.4% on MMLU-Pro. Ranked 43rd in the world. It runs on Australian servers. Nothing ever leaves Australia, and nothing you type is used to train AI. Delete a chat and it’s gone for good. Most of the world’s AI answers to somebody else. This one answers only to you. Free to start: agentaus.com.au
1
3
1
28
834
Fenix retweeted
Así se hacía antes de que existiera la IA
We built Higgsfield Miracle to address one of the biggest challenges in AI video: editing a specific part of a shot without changing the rest. Powered by GPT-6.2 Astra, it splits any video into layers for pixel-level editing in real time, with world knowledge built in. Every shot keeps the image depth that cinema-grade post-production needs. Now in private beta with select testers. Coming soon to Higgsfield MCP. Set up Higgsfield plugin in ChatGPT to be the first to try it.
10
12
68
12,555
Fenix retweeted
PUEDES MONTAR UN NEGOCIO DE +10K/MES CON ESTA IA Y CASI NADIE LO ESTÁ HACIENDO Reducto convierte cualquier PDF o foto de un documento en datos limpios. Lo compararon con 8 parsers en 12 benchmarks y ganó la mayoría Los autónomos odian pelearse con sus facturas cada trimestre. Tú se lo automatizas: 1. Te mandan sus facturas por WhatsApp o email 2. Reducto saca proveedor, fecha, base, IVA y total 3. Claude lo valida y lo clasifica 4. Se exporta solo a su Excel o programa contable Los números: → 110 autónomos a $99/mes = $10.890/mes
→ Costes: ~$400/mes
→ Margen: ~96% Guárdalo y empieza hoy
Last week, I said every parser is #1 on its own benchmark. So we tested the benchmarks. We ran 8 parsers across 12 benchmarks, then tilted the pages by 2–4°. 6 benchmarks got a different winner. We’re calling it BenchBench.
15
11
45
6,681
Fenix retweeted
usuarios de chatgpt viendo esto
🤝 Paid partnership
We built Higgsfield Miracle to address one of the biggest challenges in AI video: editing a specific part of a shot without changing the rest. Powered by GPT-6.2 Astra, it splits any video into layers for pixel-level editing in real time, with world knowledge built in. Every shot keeps the image depth that cinema-grade post-production needs. Now in private beta with select testers. Coming soon to Higgsfield MCP. Set up Higgsfield plugin in ChatGPT to be the first to try it.
19
20
54
6,509