Matías Matthews

17 posts

Matías Matthews banner
Matías Matthews

Matías Matthews

@MatiasMatthews

Founder en Valaris. AI operations for ambitious teams. Escribo sobre IA aplicada, startups y ejecución en LatAm.

Santiago, Chile Katılım Ağustos 2009
135 Takip Edilen16 Takipçiler
Sabitlenmiş Tweet
Matías Matthews
Matías Matthews@MatiasMatthews·
Estoy empezando a usar X para compartir lo que aprendo construyendo Valaris. Me interesan 3 cosas: 1. IA aplicada a operaciones reales 2. Startups y ejecucin en LatAm 3. Cómo convertir tecnologa en sistemas que funcionen Voy a escribir sobre lo que veo, pruebo y construyo.
Español
0
0
0
86
Garry Tan
Garry Tan@garrytan·
Founders must stop trying to building 2010-era businesses with 2026-era technology. Don't try to rebuild Foursquare or Yelp. Don't try to recreate Basecamp by 37 Signals with $10/mo SaaS pricing. Don't underprice! If it works it's worth a lot more. Don't be tempted to become "Tech enabled PE" with revenue tricks. The rules of tech changed with AI. Play the new game.
English
305
172
2.6K
176.8K
Matías Matthews
Matías Matthews@MatiasMatthews·
@garrytan 100%. A lot of 2010-era SaaS assumed humans would do the orchestration and software would store the workflow. With AI, the interesting shift is that software can start carrying the operational judgment too: context, exceptions, prioritization, follow-up. New game indeed.
English
0
0
0
3
Matías Matthews
Matías Matthews@MatiasMatthews·
La mayoría de las conversaciones sobre IA parten desde el modelo. En la práctica, muchas veces el problema está antes: información dispersa, contexto incompleto, decisiones que dependen de tres personas, criterios que viven en la cabeza de alguien, procesos que solo funcionan cuando “la persona clave” está conectada. En LatAm eso se nota más, porque operamos con más fricción y menos holgura. Por eso me interesa menos la IA como magia y más la IA como una nueva forma de ordenar criterio.
Español
1
0
1
39
Matías Matthews
Matías Matthews@MatiasMatthews·
Mi socio Seba lleva casi 30 años programando. El otro día me dijo algo que me dejó con cara de niño mirando un cohete: “Mati, llevo como 6 meses sin programar. Solo orquestar, planificar y revisar LLMs.” Y lo dijo fascinado, no asustado. Quizás el oficio no se está muriendo. Está subiendo de capa.
Español
0
0
0
26
Alfonso de los Ríos
Alfonso de los Ríos@ponchodelosrios·
Programo desde que tengo 11 años, y este fin de semana volando a San Francisco, escuché una entrevista a Jansen Huang en Stanford sobre una crisis existencial: Ahora, que hacemos los programadores? Buena conversación sobre lo que sigue para los programadores, después de que por 64 años no ha cambiado nada en la manera en la que programamos. - "No optimicen solo para la pasión. Busquen también el sufrimiento productivo." - El 90% del trabajo de CEO es duro. Solo el 10% es divertido. Pero el sufrimiento construye el músculo de resiliencia que vas a necesitar cuando tu equipo, tu familia, o tu empresa dependa de que tú aguantes. - El stack entero se está reconstruyendo. Escoge una capa. Reinventala. Lo escuché y pensé en cada uno de los momentos en Nowports y Handle donde ese músculo fue la diferencia entre seguir o cerrar. youtube.com/watch?v=tsQB0n…
YouTube video
YouTube
Español
1
1
30
2.2K
Harj Taggar
Harj Taggar@harjtaggar·
I don’t think local models will ever be the default but every device will need one as an emergency generator to stop people freaking out when they lose connectivity and have to make a decision.
English
48
23
232
79.6K
Matías Matthews
Matías Matthews@MatiasMatthews·
@harjtaggar @ycombinator Local models as the flashlight in the drawer. Not the default, but absolutely necessary when the lights go out. The interesting question is: which decisions must still work offline?
English
0
0
0
255
Matías Matthews
Matías Matthews@MatiasMatthews·
@vkhosla This is the part of AI that still feels a little magical to me: it makes the learning curve feel young again. Maybe the real edge is curiosity stamina.
English
0
0
0
5
Vinod Khosla
Vinod Khosla@vkhosla·
At 70+ I feel like I'm thirty backing exciting innovators/innovations making the impossible possible and from improbable ideas to world-changing socially impactful companies. Never had such a fast learning rate in my life or felt I could have so much impact, which is exciting.
English
160
132
1.5K
435.8K
Matías Matthews
Matías Matthews@MatiasMatthews·
@amasad I love this framing. If software is creative work now, the hard part isn’t just generating code; it’s designing the loop where humans bring taste, constraints and judgment, while agents do the heavy lifting. Feels like we’re all learning a new instrument.
English
0
0
0
6
Amjad Masad
Amjad Masad@amasad·
Software isn’t merely technical work anymore. It’s creative. Introducing Replit Agent 4. The first AI built for creative collaboration between humans and agents. Design on an infinite canvas, work with your team, run parallel agents, and ship working apps, sites, slides & more.
English
665
691
6.9K
3.1M
Matías Matthews
Matías Matthews@MatiasMatthews·
@saranormous This is one of those tiny operational black holes that quietly eats teams. The funny bit: the build vs buy debate often forgets the third option: compose the boring parts, and only invent where the workflow is genuinely weird. I keep relearning this.
English
0
0
0
11
sarah guo
sarah guo@saranormous·
😓 the number of companies out there that won't pay $20K a year for off-the-shelf SaaS, but then spend unlimited engineering effort building internal tooling themselves and then complain they don't have enough engineering capacity
English
92
305
3.7K
0
Matías Matthews
Matías Matthews@MatiasMatthews·
@AnjneyMidha Yes. The chart is often the first model the human builds in their head. If that model is messy, no benchmark number can save the conversation.
English
0
0
1
227
Anjney Midha
Anjney Midha@AnjneyMidha·
if you run an ai lab, pls ensure your team has read this before putting any charts out into the world
Anjney Midha tweet media
English
55
123
2K
1.3M
Dan Nunn
Dan Nunn@danyay·
Curious how you use skills to get new chats up to speed on repos? Or do you use huge compacted threads? I keep all the details in a Markdown file I reference (deploy details, product intent, etc), but feel like there’s a missing opportunity to use a skill to train agents about repos (esp when I have to work with multiple repos for a feature)
English
12
1
16
215.2K
@levelsio
@levelsio@levelsio·
My laptop screen looks pretty much like this all day now and same on my iPhone when I work (which could be anywhere now) It's just tabs for my sites, all on a VPS, synced with my iPhone via @TermiusHQ (unaffiliated) and usually with Claude Code open to fix or build new things
@levelsio tweet media@levelsio tweet media@levelsio tweet media
@levelsio@levelsio

So here's my latest set up Every site I have is a profile on Termius like > hoodmaps .com I click it and immediately I'm in my server and I get dropped in a tmux session that's always tied to the corresponding site I wanna log in to To make this work I have this startup snippet in each site's Termius profile: > cd /srv/http/hoodmaps.com && tm (so /srv/http is where my sites are and then hoodmaps .com is the example site here, and "&& tm" is the important part here) Then in my ~/.bashrc file I added this (written by Claude Code) which defines the "tm" function, again all it does it just put me in the right tmux session based on the folder I'm in The result is I can switch without interruption from my laptop to phone in Termius with auto reconnecting sessions and usually I just have Claude Code open in each session to work Before I had to mess around with 1) not having smooth switching from laptop to phone, I'd have to use Claude Code's /resume for it, annoying, 2) having multiple sessions for same sites, gets messy and confusing fast, now it FORCES me into one session per site, this just works so well, I'm so fast, and each of my sites is just an open tab in Termius, I've never worked so structured and clean! Here is the code, maybe it helps somebody: # tmux session per folder. `tm` (no args) attaches to / creates a session # named after the current dir's basename. `tm name` overrides the name. # Works whether already inside tmux (uses switch-client) or outside it. tm() { command -v tmux >/dev/null 2>&1 || { echo "tmux not installed"; return 1; } local name="${1:-$(basename "$PWD")}" # tmux session names can't contain '.' or ':' — replace with '-' name="${name//./-}" name="${name//:/-}" if [ -n "$TMUX" ]; then tmux has-session -t "$name" 2>/dev/null || tmux new-session -d -s "$name" -c "$PWD" tmux switch-client -t "$name" else tmux attach -t "$name" 2>/dev/null || tmux new -s "$name" -c "$PWD" fi } # Auto-attach on interactive login: picks a session named after wherever # you land. Plain `ssh server` lands in $HOME → session "root". Use # `ssh server -t "cd /srv/sm.levels.io && bash -l"` to land in a site # folder → session "sm-levels-io". Skips inside tmux and non-interactive # shells so scp/rsync/scripted ssh keep working. if command -v tmux >/dev/null 2>&1 && [ -z "$TMUX" ] && [[ $- == *i* ]]; then tm fi

English
192
129
3K
831.1K
Matías Matthews
Matías Matthews@MatiasMatthews·
@danshipper @every Automation eats the obvious work. Then suddenly taste, judgment, and weird human questions become the bottleneck.
English
0
0
0
18
Dan Shipper 📧
Dan Shipper 📧@danshipper·
We’ve automated every single thing we can @every with AI agents. And yet there’s way more human work to do than ever. We’ve gone from 4 -> 30 human employees since GPT-3. I wrote a report on the structural reasons: how AI makes expert competence cheap, why that drives up demand for experts, and why the dynamic only intensifies as we approach AGI. After Automation: every.to/p/after-automa…
Dan Shipper 📧 tweet media
English
162
278
2.2K
1.3M
Matías Matthews
Matías Matthews@MatiasMatthews·
@GoogleDeepMind Multimodality gets really interesting the moment the brief survives the handoff. Same context in, many artifacts out.
English
0
0
0
4
Google DeepMind
Google DeepMind@GoogleDeepMind·
We’re dropping Gemini Omni: our first step towards a model that can create anything from anything - starting with video. It combines Gemini’s intelligence with our generative media systems - representing a leap forward in world understanding, multimodality, and editing 🧵
English
411
1.3K
8.4K
1.2M
Matías Matthews
Matías Matthews@MatiasMatthews·
@ycombinator Feels like compute is becoming a financing instrument, not just infrastructure. The real question isn’t ‘would you take the $2M?’ It’s whether your team has the operating discipline to turn credits into learning loops before they turn into burn with prettier graphs.
English
0
1
3
1K
Y Combinator
Y Combinator@ycombinator·
OpenAI is offering $2M in tokens to every YC company in the spring and summer batches. We extended the summer deadline to May 25 so more founders can get in on it. ycombinator.com/apply
English
106
98
1.1K
752.1K
Matías Matthews
Matías Matthews@MatiasMatthews·
La próxima ola de IA en LatAm no la va a ganar el mejor modelo. La va a ganar quien cierre mejor este loop: capturar señal ejecutar trabajo medir resultado volver a aprender Si un agente no mueve operación, margen o tiempo de respuesta, sigue siendo una demo cara. Estamos entrando en la etapa donde importa menos “wow” y más “run rate”. Y, honestamente, ahí se pone bueno.
Español
0
0
0
24