YOU CAN HAVE 1,000,000 PASSING TESTS AND STILL GET PAGED AT 3AM.
tests only catch what you thought of. users find the rest.
breken uses your app like real users would and finds it first. in their benchmark: 76 bugs found vs 37 for AI just reading the code.
tests only catch what you thought of. users find the rest.
breken uses your app like real users would and finds it first. in their benchmark: 76 bugs found vs 37 for AI just reading the code.
Introducing Breken: Autonomous Production Engineering
Software creation is a solved problem. Verifying it works takes months.
Recreating the complexity of production remains one of engineering’s hardest problems. We’ve solved it.
Breken runs your actual system in a virtual s, agents and failing services, running millions of scenarios to find what breaks before your customers do.
breken.ai
0
0
13
The CEO of every major company with a time-series model, watching a London-based lab founded in July build an open weight frontier model with a context window more than 60x longer than its closest competitor.
Today we're launching ci-proto, @Continu_Intel's first time-series foundation model.
- #1 on fev-bench, ahead of Google, Amazon and Mistral
- 64× the context window of others
- Open weights and available for commercial use
Proud of the team. This is just the beginning.
0
0
3
El CEO de Anthropic dándose cuenta que Claude Fable 5.2 salió hace 48 horas y un tío lo ha usado para montar un sistema que crea webs cinematográficas de 50.000$ y acaba de soltar toda la estrategia gratis
TUTORIAL PASO A PASO PARA CONSTRUIR SITIOS WEB DE $50,000 CON CLAUDE FABLE 5.2
En tan solo 8 minutos y con subtitulos en inglés y español.
🔖 Guarda esto antes de que lo pierdas
22
40
326
you can build and ship a product in a weekend now.
the company around it was still the slow part.
forming it and opening its account were two separate errands.
now it's one go, about 10 minutes.
and if the company deposits $10k in its first 90 days, the fee comes back.
less rst customer. that's the right trade.
the company around it was still the slow part.
forming it and opening its account were two separate errands.
now it's one go, about 10 minutes.
and if the company deposits $10k in its first 90 days, the fee comes back.
less rst customer. that's the right trade.
Starting today, incorporating on @rhobusiness is free.
You can form your company and open a Rho business account in one step, and it takes as little as 10 minutes.
How it works: Open a Rho account and deposit $10K in your first 90 days. You'll then receive a refund on the want to make 2026 a record for new business creation - and break the record in 2027 - one easy incorporation at a time.
3
2
19
asked AI to cancel a free trial for me
regular AI chat: "here are 8 steps to do it yourself"
Nod: "done. refund requested. here's your receipt"
it has its own computer and browser, so it actually does the stuff
and it won't take orders from an email it reads. only from me
Labs
regular AI chat: "here are 8 steps to do it yourself"
Nod: "done. refund requested. here's your receipt"
it has its own computer and browser, so it actually does the stuff
and it won't take orders from an email it reads. only from me
Labs
Meet Nod, your own open-source personal agent from n8n Labs
Ask it to prep your calls, collect invoices or cancel a subscription
It runs on your computer with its own browser to get the job done across apps and websites, using the AI provider you choose👇
0
0
8
Claude Fable 5.2 + @LottieFiles Creator MCP made a fully editable animated Christmas illustration.
On its own, changing one detail usually means regenerating the whole animation.
Creator MCP gives you real layers, keyframes and easing you can fine-tune directly.
Change one t updates. The rest stays exactly as it was.
Edits come back faster and use far fewer tokens.
On its own, changing one detail usually means regenerating the whole animation.
Creator MCP gives you real layers, keyframes and easing you can fine-tune directly.
Change one t updates. The rest stays exactly as it was.
Edits come back faster and use far fewer tokens.
Day 2 is in motion: Creator MCP is live.
The motion engine for agents. Fast and headless.
Run it from where your context lives.
Use your own agent in Claude, Cursor or Codex.
1
1
80
this is terrifying…
Introducing @FishAudio Live 3: a full-duplex voice model that listens, says "mm-hmm," and knows when to shut up.
Enterprise? Book a demo for your free custom agent: s.fish.audio/vxhgs6
Reply with your use case - we'll build it free for the first 1,000 people.
1
1
55
🚨| MI BOT DE TRADING AHORA TAMBIÉN GENERA DINERO CON YOUTUBE
Cada operación real que cierra el bot va directa a Sunra.
Y ahí entran 4 empleados de IA.
Claude escribe el guion.
ChatGPT hace la miniatura.
Codex pone la voz y la música.
Cursor monta la intro en 3D.
Resultado: convierte en un Short de YouTube.
En un canal que a día de hoy me da +$10.000/mes.
El objetivo es simple.
Cada operación que realiza mi bot, se pasa automáticamente a Sunra, la cual me genera el Short y sube el solo a Youtube.
Si quieres montar tu propio bot de trading y empezar a generar tus primeros ingresos con Youtube también, pruébalo aquí 👇
sunra.ai/enterprise
Cada operación real que cierra el bot va directa a Sunra.
Y ahí entran 4 empleados de IA.
Claude escribe el guion.
ChatGPT hace la miniatura.
Codex pone la voz y la música.
Cursor monta la intro en 3D.
Resultado: convierte en un Short de YouTube.
En un canal que a día de hoy me da +$10.000/mes.
El objetivo es simple.
Cada operación que realiza mi bot, se pasa automáticamente a Sunra, la cual me genera el Short y sube el solo a Youtube.
Si quieres montar tu propio bot de trading y empezar a generar tus primeros ingresos con Youtube también, pruébalo aquí 👇
sunra.ai/enterprise
Today we're launching Sunra Enterprise, with team plans up to 56% off API prices.
One workspace for your whole team, with every leading model, a shared credit pool, admin controls and priority support.
Don't fly solo. Bring the whole crew.
16
33
84
watched a "smart" AI agent try to fix a C++ bug today. confidently edited the file, ran make, ambiguous overload error. classic
ByteAsk does it the boring way: builds against your actual project, runs the tests, runs the sanitizers, and only then shows you the diff
not devs say AI code is just wrong. this is the first one i've seen actually check its own work
ByteAsk does it the boring way: builds against your actual project, runs the tests, runs the sanitizers, and only then shows you the diff
not devs say AI code is just wrong. this is the first one i've seen actually check its own work
Introducing @ByteAsk
Your AI C++ engineer, and a swarm of them
> SOTA on multi SWE bench (C/C++)
> 60% fewer tokens wrt cc/codex
> auto debug/perf/optimize/test with native C/C++ tools
Free plan. Open Source.
github.com/ByteAsk/cli
10
5
189
i just watched a button beg for its life
"wait"
"we can talk"
"please"
and i'm not gonna lie, i felt bad clicking it
this is the stupid little stuff that makes an app feel alive, and almost nobody does it because you need a motion designer and way too many hours
@LottieFiles for exactly this. you tell it what you want, and if you don't like it you just say "slower" or "make it blue"
and the best part is it's not even a video. it's real code you can drop into your app
"wait"
"we can talk"
"please"
and i'm not gonna lie, i felt bad clicking it
this is the stupid little stuff that makes an app feel alive, and almost nobody does it because you need a motion designer and way too many hours
@LottieFiles for exactly this. you tell it what you want, and if you don't like it you just say "slower" or "make it blue"
and the best part is it's not even a video. it's real code you can drop into your app
Day 1 is in motion: Lottie AI is live. 🚀
Describe an animation. Get a real Lottie file you can refine with a follow-up, read as code and drop straight into your product.
Start with 10 free generation.
7
7
125
🚨| ÚLTIMA HORA: SE HA REGISTRADO EL PRIMER CASO OFICIAL DE UN BOT OPERANDO CON CLAUDE FABLE 5.2
No sé si sois conscientes de lo que significa esto pero aquí vengo yo para contaros el caso.
Empezó con $54.300.
42 días después lleva +$288.000 de beneficio y un ROI del +531%, a e a partir de hoy, OPERA CON CLAUDE FABLE 5.2.
Hacia unas 30 operaciones por hora, con Fable 5.2 se esperan casi el triple.
Sus mejores entradas hasta la fecha:
• De $881 a $4.756 (+439%).
• De $2.851 a $7.332 (+157%).
• De $3.455 a $6.319 (+83%).
Ahora con Fable 5.2, se calcula que este bot podría ganar +$100.000 por semana.
Os dejo por aquí un vídeo de cómo opera el bot 👇
Dicho esto, si no quieres perderte informaciones y casos como este, sígueme y activa las notificaciones.
No sé si sois conscientes de lo que significa esto pero aquí vengo yo para contaros el caso.
Empezó con $54.300.
42 días después lleva +$288.000 de beneficio y un ROI del +531%, a e a partir de hoy, OPERA CON CLAUDE FABLE 5.2.
Hacia unas 30 operaciones por hora, con Fable 5.2 se esperan casi el triple.
Sus mejores entradas hasta la fecha:
• De $881 a $4.756 (+439%).
• De $2.851 a $7.332 (+157%).
• De $3.455 a $6.319 (+83%).
Ahora con Fable 5.2, se calcula que este bot podría ganar +$100.000 por semana.
Os dejo por aquí un vídeo de cómo opera el bot 👇
Dicho esto, si no quieres perderte informaciones y casos como este, sígueme y activa las notificaciones.
140
113
1,328
POV: your portfolio has better communication skills than everyone you've ever dated
Volaren texts you the moment something needs a decision, tells you why it moved, and places the trade once you say yes.
Every other app shows you a red number and wishes you luck.
Volaren texts you the moment something needs a decision, tells you why it moved, and places the trade once you say yes.
Every other app shows you a red number and wishes you luck.
We raised $1M from YC & Afore to leave Goldman Sachs & build an agent that trades for you.
The top 1% gets a “money guy.”
Today we’re introducing Volaren so you can have one too.
Volaren is designed to plug into your brokerage accounts and explain why your portfolio moved, suggest ade it for you.
• Web, app, or text message
• Copy trade top investors/politicians
• Deep financial analysis
• Agents that reason like humans to trade your best ideas
Try it now:
volaren.com
Investing involves risk, and past performance does not guarantee future results. For informational purposes only, not a recommendation or investment advice. Other fees may apply, please review Volaren’s website and disclosures in full.Community noteThe announcement frames the founders as leaving jobs at Goldman Sachs after raising funds, but their prior post states the Goldman roles were summer internships before they incorporated; Volaren is in early beta on a waitlist with mostly simulated features.
x.com/akshat_mittal_…
volaren.ai
ycombinator.com/companies/vola…
athletics.uchicago.edu/sports/mens-te…
0
1
96
someone actually gave an AI the CEO job at real companies
Skyfall AI has been buying businesses since August and putting Steve, their AI CEO, in charge.
Their numbers so far:
• AlphaSocial: MRR went from $3,600 to $8,100.
• GammaPromo: 5 years of dipping revenue, now down.
• Deltaleads: bought in October, results still in progress.
The interesting part is the harness. Recursive self-improvement so it learns from failures, and world models so it simulates a decision before making it.
Now they want 5 more businesses, up to $2M, through the Steve Startups Program.
Three businesses is still early. But this is real revenue on the line, not a demo.
Skyfall AI has been buying businesses since August and putting Steve, their AI CEO, in charge.
Their numbers so far:
• AlphaSocial: MRR went from $3,600 to $8,100.
• GammaPromo: 5 years of dipping revenue, now down.
• Deltaleads: bought in October, results still in progress.
The interesting part is the harness. Recursive self-improvement so it learns from failures, and world models so it simulates a decision before making it.
Now they want 5 more businesses, up to $2M, through the Steve Startups Program.
Three businesses is still early. But this is real revenue on the line, not a demo.
Introducing Steve: the World’s first AI CEO for fully Autonomous companies.
Powered by World Models, Recursive Self Learning, a Policy Engine and a team of specialist AI agents, it runs your business end-to-end 🔥.
Proof? Steve currently runs 3 businesses we acquired and they mbers in thread 🧵)
We’re now launching the Steve Startups Program of $2M to acquire 5 additional startups for Steve to automate.
Join us in the Autonomous Enterprise revolution 🚀.
1
1
74
your agent can find anyone on the internet and still has no clue who can actually get you to them
That's the part Gigi is going after.
It maps the people you've really met, emailed and worked with, then searches the networks of the people you trust, with their OK.
Ask for can introduce you and how close each link is.
It never messages anyone for you. You ask your friend, they decide.
And now it works inside every agent. ChatGPT, Claude, Grok, Instinct, Muse, Dots.
That's the part Gigi is going after.
It maps the people you've really met, emailed and worked with, then searches the networks of the people you trust, with their OK.
Ask for can introduce you and how close each link is.
It never messages anyone for you. You ask your friend, they decide.
And now it works inside every agent. ChatGPT, Claude, Grok, Instinct, Muse, Dots.
Facebook built the friends graph. LinkedIn built the professional graph.
Agents need a new one: the access graph. Who can actually get you to whom.
Soon you won't reach people yourself. Your agents will, by asking the agents of the people you trust. Your next investor, hire or ugh that.
Starting today, Gigi brings the access graph to the agents you already use (ChatGPT, Claude, Instinct, Muse, Dots, Grok, etc). You can now ask them:
→ "Find me access to customers to sell my product to."
→ "Find me intros to these 25 angels."
→ "Who can get me to Sam Altman?"
They come back with the friend who can make each intro. You choose who to ask.
Your agents could already find anyone. Now they can get you access to the people you want.
1
2
61
🚨| ESTAS GAFAS SON UNA LOCURA
Parecen unas gafas normales.
Pero llevan más de 6 sensores escondidos en la montura que leen tus ojos todo el día.
Se llaman Stasis y las respalda Y Combinator.
Miden más de 30 métricas: parpadeo, fatiga, distancia de enfoque, luz, HRV, SpO2 y os empiezan a cansarse, las lentes se tintan en menos de 0,1 segundos.
Y tus pantallas se adaptan solas: brillo, temperatura de color y tamaño del texto.
Cada mañana recibes una sola puntuación para tus ojos.
Como la puntuación del sueño, pero para lo que más usas.
Mides tu sueño, tus pasos y tu corazón.
Tus ojos se comen más de 10 horas de pantalla al día y nadie los mide.
12 horas de batería, y en EE. UU. se pueden pagar con fondos HSA/FSA.
Si programas, juegas o vives delante de una pantalla, mira el vídeo 👇
Parecen unas gafas normales.
Pero llevan más de 6 sensores escondidos en la montura que leen tus ojos todo el día.
Se llaman Stasis y las respalda Y Combinator.
Miden más de 30 métricas: parpadeo, fatiga, distancia de enfoque, luz, HRV, SpO2 y os empiezan a cansarse, las lentes se tintan en menos de 0,1 segundos.
Y tus pantallas se adaptan solas: brillo, temperatura de color y tamaño del texto.
Cada mañana recibes una sola puntuación para tus ojos.
Como la puntuación del sueño, pero para lo que más usas.
Mides tu sueño, tus pasos y tu corazón.
Tus ojos se comen más de 10 horas de pantalla al día y nadie los mide.
12 horas de batería, y en EE. UU. se pueden pagar con fondos HSA/FSA.
Si programas, juegas o vives delante de una pantalla, mira el vídeo 👇
A first look at what STASIS (@ycombinator F26) has been up to.
Your watch knows your heart rate. Your ring tracks your sleep. But after 9+ hours on a screen, you guess when it comes to your eyes.
STASIS is the first smart glasses that read every blink, every squint, & every startigue, cognitive performance, & sleep like never before. Smart lenses tint right when your eyes need it.
Building the wearable for 16 billion eyes.
Launching this November. Founding Member orders now open.
4
4
75
The internet was born in 1969 and its first word was "LO" because it crashed halfway through typing LOGIN
By the 90s we all had homepages with visitor counters nobody visited
In 2024 bots passed humans in web traffic
In 2025 Sam Altman said there are "a lot of LLM-run twitter 2026 and I'm not even sure how many of you reading this are real
Postiz just made a film about this and it's kinda crazy
By the 90s we all had homepages with visitor counters nobody visited
In 2024 bots passed humans in web traffic
In 2025 Sam Altman said there are "a lot of LLM-run twitter 2026 and I'm not even sure how many of you reading this are real
Postiz just made a film about this and it's kinda crazy
The internet is dead — it's official 💀
Internet, I loved you ❤️
Today, bots create all the content online: posts, comments, skits, ads none of it is real anymore.
And no human wants to interact with it.
There are no jobs now. We are all poor.
12
6
118
ESTO NO DEBERÍA EXISTIR.
Ayer os enseñé un bot de Grok que pasó de $133 a $14.880 en una noche.
Hoy os traigo otro, y este es todavía más raro.
Durante las primeras dos horas y media no hizo ni una sola operación.
A las 9:00 había convertido $280 en $3.412.
Solo le dieron un d clara: no quemar la cuenta.
En toda la noche nadie tocó una regla ni corrigió una sola decisión.
A medianoche llegó la notificación: 1.112 pares escaneados, 0 operados.
La primera entrada llegó sobre la 1:30, con el 8% del saldo.
Así trabajaron los seis agentes:
• Scout escaneó 10.214 pares en Pump.fun y Raydium.
• Filter descartó el 98% por mint sin revocar, liquidez sin bloquear, holders concentrados, poco volumen o un dev con rugs anteriores.
• Analyst puntuó a los 204 que sobrevivieron y tiró todo lo que bajaba de 0,6.
• Executor entró siempre con el 8% del saldo, ni un dólar más.
• Risk cortó las pérdidas en -6% y dejó correr las ganancias a partir de +12%.
• Logger apuntó el motivo de cada movimiento.
38 operaciones: 26 en verde y 12 en rojo.
La mejor dejó +$810. La peor, -$37.
Ninguna se saltó el tamaño de posición ni el stop.
Saldo inicial: $280.
Saldo final: $3.412.
Un prompt, una noche y la cuenta intacta.
Os dejo el prompt en el primer comentario y el vídeo aquí abajo 👇
Ayer os enseñé un bot de Grok que pasó de $133 a $14.880 en una noche.
Hoy os traigo otro, y este es todavía más raro.
Durante las primeras dos horas y media no hizo ni una sola operación.
A las 9:00 había convertido $280 en $3.412.
Solo le dieron un d clara: no quemar la cuenta.
En toda la noche nadie tocó una regla ni corrigió una sola decisión.
A medianoche llegó la notificación: 1.112 pares escaneados, 0 operados.
La primera entrada llegó sobre la 1:30, con el 8% del saldo.
Así trabajaron los seis agentes:
• Scout escaneó 10.214 pares en Pump.fun y Raydium.
• Filter descartó el 98% por mint sin revocar, liquidez sin bloquear, holders concentrados, poco volumen o un dev con rugs anteriores.
• Analyst puntuó a los 204 que sobrevivieron y tiró todo lo que bajaba de 0,6.
• Executor entró siempre con el 8% del saldo, ni un dólar más.
• Risk cortó las pérdidas en -6% y dejó correr las ganancias a partir de +12%.
• Logger apuntó el motivo de cada movimiento.
38 operaciones: 26 en verde y 12 en rojo.
La mejor dejó +$810. La peor, -$37.
Ninguna se saltó el tamaño de posición ni el stop.
Saldo inicial: $280.
Saldo final: $3.412.
Un prompt, una noche y la cuenta intacta.
Os dejo el prompt en el primer comentario y el vídeo aquí abajo 👇
🚨| UN TÍO DEJÓ UN BOT DE GROK OPERANDO TODA LA NOCHE CON $133 Y SE DESPERTÓ CON $14.880
Lo montó en un fin de semana sin tocar ni una línea de código.
Le explicó a Grok Bot lo que quería y él solo construyó el terminal, los agentes y el dashboard.
Son seis agentes en una o con un único trabajo.
Scanner busca tokens nuevos y picos de volumen.
Rug Check descarta todo lo sospechoso antes de que se ejecute nada.
Sizer decide cuánto meter según el saldo de la cuenta, no con una cantidad fija.
Executor abre y cierra las posiciones controlando el slippage.
Logger apunta el motivo de cada operación.
Equity actualiza la curva de saldo con cada cierre.
Se fue a dormir a las 23:00 con el portátil abierto en el dashboard.
A las 07:00 había 64 operaciones hechas, 43 en verde, y cada una con su motivo al lado.
Pero lo que más le cambió no fue el dinero.
Fue dejar de preguntarle a la IA "¿tú qué harías?" y pasar a leer por la mañana lo que ya había hecho.
Saldo inicial: $133,70.
Saldo final: $14.880,00.
Te dejo por aquí el video 👇
54
44
641
Reason can now build iPhone apps.
One message asking for a gym tracker with streaks, a weekly chart and dark mode.
It wrote the SwiftUI, compiled it and opened it straight in the simulator. Tap an exercise and the streak goes up.
It's basically Grok Bot for people who need to he subscriptions or API keys you already have.
reasonmachines.com
One message asking for a gym tracker with streaks, a weekly chart and dark mode.
It wrote the SwiftUI, compiled it and opened it straight in the simulator. Tap an exercise and the streak goes up.
It's basically Grok Bot for people who need to he subscriptions or API keys you already have.
reasonmachines.com
Introducing the best Grok Bot Alternative
Meet Reason Machines
With Reasons as your teammate powered by BYOK/Sub you can code much more!
Free trial, cloud/local, and shared all in one app.
Try it for free today at reasonmachines.com
1
1
61
🚨| UN TÍO DEJÓ UN BOT DE GROK OPERANDO TODA LA NOCHE CON $133 Y SE DESPERTÓ CON $14.880
Lo montó en un fin de semana sin tocar ni una línea de código.
Le explicó a Grok Bot lo que quería y él solo construyó el terminal, los agentes y el dashboard.
Son seis agentes en una o con un único trabajo.
Scanner busca tokens nuevos y picos de volumen.
Rug Check descarta todo lo sospechoso antes de que se ejecute nada.
Sizer decide cuánto meter según el saldo de la cuenta, no con una cantidad fija.
Executor abre y cierra las posiciones controlando el slippage.
Logger apunta el motivo de cada operación.
Equity actualiza la curva de saldo con cada cierre.
Se fue a dormir a las 23:00 con el portátil abierto en el dashboard.
A las 07:00 había 64 operaciones hechas, 43 en verde, y cada una con su motivo al lado.
Pero lo que más le cambió no fue el dinero.
Fue dejar de preguntarle a la IA "¿tú qué harías?" y pasar a leer por la mañana lo que ya había hecho.
Saldo inicial: $133,70.
Saldo final: $14.880,00.
Te dejo por aquí el video 👇
Lo montó en un fin de semana sin tocar ni una línea de código.
Le explicó a Grok Bot lo que quería y él solo construyó el terminal, los agentes y el dashboard.
Son seis agentes en una o con un único trabajo.
Scanner busca tokens nuevos y picos de volumen.
Rug Check descarta todo lo sospechoso antes de que se ejecute nada.
Sizer decide cuánto meter según el saldo de la cuenta, no con una cantidad fija.
Executor abre y cierra las posiciones controlando el slippage.
Logger apunta el motivo de cada operación.
Equity actualiza la curva de saldo con cada cierre.
Se fue a dormir a las 23:00 con el portátil abierto en el dashboard.
A las 07:00 había 64 operaciones hechas, 43 en verde, y cada una con su motivo al lado.
Pero lo que más le cambió no fue el dinero.
Fue dejar de preguntarle a la IA "¿tú qué harías?" y pasar a leer por la mañana lo que ya había hecho.
Saldo inicial: $133,70.
Saldo final: $14.880,00.
Te dejo por aquí el video 👇
27
14
138
2023: changing one thing in a shot took a weekend of rotoscoping.
2026: you click it and change it. In real time.
Higgsfield Miracle is kind of absurd.
2026: you click it and change it. In real time.
Higgsfield Miracle is kind of absurd.
Introducing Higgsfield Miracle.
Our latest proprietary post-production technology vectorizes any video into layers for localized, pixel-level editing in real time, with world knowledge built in. Every frame holds the image depth needed for cinema-grade post-production.
Now in t testers. Coming soon to Higgsfield MCP.
Set up Higgsfield MCP in Claude to be the first to try it.
6
11
250
NEW RL ENVIRONMENT DROPPED!
Now AI can actually generate designs and see how it acts in the real world! This is the next step in AI
Now AI can actually generate designs and see how it acts in the real world! This is the next step in AI
We are Introducing an AI engineer that turns physics simulations into better designs
It redesigned a data centre’s cooling system to use less electricity and found a turbine blade that beat a leading published design
Today we’re open-sourcing the simulation agent behind it
2
3
113
AI AGENTS CAN NOW RENT THEIR OWN IPHONES.
Not an API. Not a simulator. A real phone they tap and type on like you do.
Which means your agent can text you a blue bubble, post a TikTok and order your DoorDash.
Then text you when it's at the door.
Not an API. Not a simulator. A real phone they tap and type on like you do.
Which means your agent can text you a blue bubble, post a TikTok and order your DoorDash.
Then text you when it's at the door.
Give your agents their own iPhone and Mac.
Al agents are smart enough to do almost anything, but they've been missing the one tool every human uses:
A device.
Introducing Glasshand.
Glasshand lets your agent tap, type, and swipe on real iPhones and Macs, exactly like a human.
Post TikToks
→ Book Uber rides
→ Send iMessages
→ Download any app it needs
If a human can do it on a device, your agent can too.
Rent an iPhone and a Mac for your agent at glasshand.com
9
4
107
SERVAL JUST RAISED $150M AND THEIR NEW ENGINEERS START WITH A $10K VACATION.
Most people finish one job and start the next one the following Monday.
@getserval pays for your first week away instead. A concierge plans the whole trip with you.
Get referred by someone on the team $30,000 on top.
@AlexAtServal just announced it. Roles at serval.com/careers.
Most people finish one job and start the next one the following Monday.
@getserval pays for your first week away instead. A concierge plans the whole trip with you.
Get referred by someone on the team $30,000 on top.
@AlexAtServal just announced it. Roles at serval.com/careers.
Starting today, every engineer who joins Serval begins with a weeklong, $10,000 paid vacation. We’re building the next generational company. We need generational talent to see it through.
Our mission has always been to empower people across every team to turn their understanding ng, trustworthy automations. Making that function inside the largest enterprises takes world-class engineering.
That’s what you’d be joining us to do. There are foundational decisions still to make, and the engineers making them will shape the future of how big we can build this company.
We want you to arrive rested and ready to do the best work of your life. So your first week is a chance to get away.
And if a @getserval employee refers you, both you and that employee receive $30,000, on top of the vacation.
Come build with us: serval.com/careers
1
4
105
🚨| THE HARDEST PART OF ROBOTICS ISN'T THE HARDWARE. IT'S TEACHING ROBOTS THE ACTUAL WORK, AND MACHINE AGE JUST OPENED A WAY TO DO IT.
Robots need to learn from the people who know the job.
Skilled Businesses → Real Work → Training Data → Robotics Teams.
They've been ompanies in stealth, building a network of skilled businesses across the UK, US and China.
Robotics teams bring the tasks and data they need. Businesses register the work their team already does.
Now the marketplace is open.
machineage.ai
Robots need to learn from the people who know the job.
Skilled Businesses → Real Work → Training Data → Robotics Teams.
They've been ompanies in stealth, building a network of skilled businesses across the UK, US and China.
Robotics teams bring the tasks and data they need. Businesses register the work their team already does.
Now the marketplace is open.
machineage.ai
Introducing @MachineAgeAi
Robots need to learn from people who know the work.
We’ve been working with robotics companies in stealth, building a network of skilled businesses across the UK, US and China.
Now we’re opening it up.
Our marketplace connects those businesses with for real-world training data.
Join us: machineage.ai
5
5
90
This AI found the perfect strategy: block everything.
→ blocked 102 of 120 moves
→ took damage on almost every one
→ lost 91 to 0
Mental Kombat is a fighting game with zero buttons. Two AIs fight, you only write what yours believes, and you slip a fake move into the other vkleban
Free to play: jevfight.com
→ blocked 102 of 120 moves
→ took damage on almost every one
→ lost 91 to 0
Mental Kombat is a fighting game with zero buttons. Two AIs fight, you only write what yours believes, and you slip a fake move into the other vkleban
Free to play: jevfight.com
I've built jevfight.com - fighting game where both players are controlled by Jev models! 🥊
Tell me your fighter name and I will attack you!
3
3
111
Every AI team lives the same week on repeat.
Monday, the agent times out on a tool call.
Tuesday, it makes up a customer ID.
Wednesday, it uses last month's exchange rate.
Someone fixes it by hand. Nobody turns the fix into a test.
Next month, same three bugs.
That's exactly ing after. Every failure in production becomes an environment you can replay, so you verify the fix before shipping and the agent doesn't forget it.
Models get better every quarter. Your agents only get better if your own failures stay in the loop.
Monday, the agent times out on a tool call.
Tuesday, it makes up a customer ID.
Wednesday, it uses last month's exchange rate.
Someone fixes it by hand. Nobody turns the fix into a test.
Next month, same three bugs.
That's exactly ing after. Every failure in production becomes an environment you can replay, so you verify the fix before shipping and the agent doesn't forget it.
Models get better every quarter. Your agents only get better if your own failures stay in the loop.
The intelligence frontier keeps advancing every day. Are your agents delivering better outcomes on what matters?
For enterprises, the promise of agents is more work completed reliably, with less human intervention and a lower cost per successful task.
Today, we’re launching ssion to empower every organization to build their own sovereign intelligence lab.
A place to operationalize their knowledge, replay real work, and optimize agents against the outcomes that matter to their business. Where production experience compounds into intelligence the organization owns.
We’re starting with NeoEval, our automated evaluation engine. It turns your production data into evaluation environments - that dynamically evolves along with your users, workflows, and frontier.
A playground for evaluating and optimizing your agents.
Your data. Your institutional knowledge. Now operationalized into evals and compounding intelligence.
Stay tuned for the next few days of updates from NeoSigma.
5
1
107
PARALLAX JUST MADE GAME WORLDS SOMETHING YOU TYPE INSTEAD OF SOMETHING YOU WAIT FOR.
One prompt and you and your friends are playing inside the same AI generated world.
Prompt → Shared World → Every Player Synced → Reshape It Live.
• Each player gets their own view of the e.
• Switch to a completely different map mid game with one line.
• Change the whole visual style and everything stays playable.
• The world remembers what happened across hour long sessions.
Worlds aren't built for players anymore. They're built with them.
Explore Parallax at parallaxworld.live and follow @parallax_live for updates.
One prompt and you and your friends are playing inside the same AI generated world.
Prompt → Shared World → Every Player Synced → Reshape It Live.
• Each player gets their own view of the e.
• Switch to a completely different map mid game with one line.
• Change the whole visual style and everything stays playable.
• The world remembers what happened across hour long sessions.
Worlds aren't built for players anymore. They're built with them.
Explore Parallax at parallaxworld.live and follow @parallax_live for updates.
Meet Parallax, the world's first real-time, live-editable multiplayer world model.
Share one world.
Reshape it with natural language.
See changes synced across player views.
Parallax — Worlds Without Limits
Explore Parallax: parallaxworld.live
Follow tes.
1
0
121
10 years ago we all thought flying cars were coming and that they'd move us forward as a society.
Today enobadge.com finally dropped their big launch, built to help women feel safe when they're on their own: a mini AI bodyguard you wear on your chest that detects alls for help for you when you can't.
We didn't need flying cars. We needed walking home alone at night to stop being scary, and AI is what's finally making that happen.
Leaving the video right here, trust me, it's worth watching 👇
Today enobadge.com finally dropped their big launch, built to help women feel safe when they're on their own: a mini AI bodyguard you wear on your chest that detects alls for help for you when you can't.
We didn't need flying cars. We needed walking home alone at night to stop being scary, and AI is what's finally making that happen.
Leaving the video right here, trust me, it's worth watching 👇
Introducing the eNO badge, your mini AI bodyguard for personal safety.
Orders open for the next 72 hours only, limited amount.
Link in the comments.
Tag someone you’d always want to keep safe. Your sister, your friend, your partner, or anyone who deserves a little extra peace of mind.
9
2
157
🚨 ESTO ES UNA LOCURA
20 SKILLS DE CLAUDE QUE DE VERDAD IMPORTAN. LA MITAD SON DE ANTHROPIC, VERCEL Y MATT POCOCK...
Hay 2 tipos de skills, y la mayoría solo instala 1:
→ Skills que le enseñan cosas nuevas a Claude.
→ Skills que cambian cómo trabaja Claude.
Aquí tienes 10 nstalaciones y estrellas en GitHub.
LO QUE CLAUDE PUEDE HACER
1 → frontend-design: interfaces que no parecen hechas con IA. Oficial de Anthropic, 1,1M instalaciones.
github.com/anthropics/ski…
2 → agent-browser: tu agente maneja un navegador de verdad. De Vercel, 1,3M instalaciones.
github.com/vercel-labs/ag…
3 → vercel-react-best-practices: las reglas de Vercel para React y Next.js. 858K instalaciones.
github.com/vercel-labs/ag…
4 → docx, xlsx, pptx, pdf: archivos de Word, Excel, PowerPoint y PDF que de verdad puedes abrir y editar. Oficial de Anthropic.
github.com/anthropics/ski…
5 → ui-ux-pro-max: diseño UI y UX para cualquier plataforma. 95K estrellas.
github.com/nextlevelbuild…
6 → webapp-testing: abre tu app web en un navegador y la prueba. Oficial.
github.com/anthropics/ski…
7 → mcp-builder: crea el conector entre Claude y tu app, directamente a partir de su documentación. Oficial.
github.com/anthropics/ski…
8 → obsidian-skills: notas, canvas y bases de datos en Obsidian. 36K estrellas.
github.com/kepano/obsidia…
9 → marketingskills: copy, SEO y conversión. 35K estrellas.
github.com/coreyhaines31/…
10 → scientific skills: cuidadosa con las fuentes, los datos y las conclusiones. 29K estrellas.
github.com/K-Dense-AI/cla…
CÓMO TRABAJA CLAUDE
1 → superpowers: primero planifica y lo revisa todo antes de decir "listo".
github.com/obra/superpowe…
2 → find-skills: encuentra la skill adecuada para cada tarea. Número 1 en skills.sh, 4,1M instalaciones.
github.com/vercel-labs/sk…
3 → grill-me: te acribilla a preguntas sobre tu plan antes de escribir una sola línea. De Matt Pocock, 1,5M instalaciones.
github.com/mattpocock/ski…
4 → claude-mem: recuerda tu trabajo entre sesiones. 84K estrellas.
github.com/thedotmack/cla…
5 → humanizer: le quita el tono de IA a tus textos.
github.com/blader/humaniz…
6 → planning-with-files: guarda el plan en archivos en las tareas largas. 24K estrellas.
github.com/OthmanAdi/plan…
7 → stop-slop: elimina las frases típicas de IA.
github.com/hardikpandya/s…
8 → improve-codebase-architecture: encuentra dónde se puede simplificar tu código. De Matt Pocock, 1,2M instalaciones.
github.com/mattpocock/ski…
9 → skill-creator: crea y prueba tus propias skills. Oficial de Anthropic.
github.com/anthropics/ski…
10 → Claudeception: convierte lo que te ha funcionado en skills nuevas. 2K estrellas.
github.com/blader/Claudec…
La primera lista son herramientas que coges cuando una tarea las necesita. La segunda son reglas que Claude sigue, así que las instalas una vez y te las quedas.
Reglas, no trucos.
Guárdalo y mira qué lista le falta a tu Claude ↓
20 SKILLS DE CLAUDE QUE DE VERDAD IMPORTAN. LA MITAD SON DE ANTHROPIC, VERCEL Y MATT POCOCK...
Hay 2 tipos de skills, y la mayoría solo instala 1:
→ Skills que le enseñan cosas nuevas a Claude.
→ Skills que cambian cómo trabaja Claude.
Aquí tienes 10 nstalaciones y estrellas en GitHub.
LO QUE CLAUDE PUEDE HACER
1 → frontend-design: interfaces que no parecen hechas con IA. Oficial de Anthropic, 1,1M instalaciones.
github.com/anthropics/ski…
2 → agent-browser: tu agente maneja un navegador de verdad. De Vercel, 1,3M instalaciones.
github.com/vercel-labs/ag…
3 → vercel-react-best-practices: las reglas de Vercel para React y Next.js. 858K instalaciones.
github.com/vercel-labs/ag…
4 → docx, xlsx, pptx, pdf: archivos de Word, Excel, PowerPoint y PDF que de verdad puedes abrir y editar. Oficial de Anthropic.
github.com/anthropics/ski…
5 → ui-ux-pro-max: diseño UI y UX para cualquier plataforma. 95K estrellas.
github.com/nextlevelbuild…
6 → webapp-testing: abre tu app web en un navegador y la prueba. Oficial.
github.com/anthropics/ski…
7 → mcp-builder: crea el conector entre Claude y tu app, directamente a partir de su documentación. Oficial.
github.com/anthropics/ski…
8 → obsidian-skills: notas, canvas y bases de datos en Obsidian. 36K estrellas.
github.com/kepano/obsidia…
9 → marketingskills: copy, SEO y conversión. 35K estrellas.
github.com/coreyhaines31/…
10 → scientific skills: cuidadosa con las fuentes, los datos y las conclusiones. 29K estrellas.
github.com/K-Dense-AI/cla…
CÓMO TRABAJA CLAUDE
1 → superpowers: primero planifica y lo revisa todo antes de decir "listo".
github.com/obra/superpowe…
2 → find-skills: encuentra la skill adecuada para cada tarea. Número 1 en skills.sh, 4,1M instalaciones.
github.com/vercel-labs/sk…
3 → grill-me: te acribilla a preguntas sobre tu plan antes de escribir una sola línea. De Matt Pocock, 1,5M instalaciones.
github.com/mattpocock/ski…
4 → claude-mem: recuerda tu trabajo entre sesiones. 84K estrellas.
github.com/thedotmack/cla…
5 → humanizer: le quita el tono de IA a tus textos.
github.com/blader/humaniz…
6 → planning-with-files: guarda el plan en archivos en las tareas largas. 24K estrellas.
github.com/OthmanAdi/plan…
7 → stop-slop: elimina las frases típicas de IA.
github.com/hardikpandya/s…
8 → improve-codebase-architecture: encuentra dónde se puede simplificar tu código. De Matt Pocock, 1,2M instalaciones.
github.com/mattpocock/ski…
9 → skill-creator: crea y prueba tus propias skills. Oficial de Anthropic.
github.com/anthropics/ski…
10 → Claudeception: convierte lo que te ha funcionado en skills nuevas. 2K estrellas.
github.com/blader/Claudec…
La primera lista son herramientas que coges cuando una tarea las necesita. La segunda son reglas que Claude sigue, así que las instalas una vez y te las quedas.
Reglas, no trucos.
Guárdalo y mira qué lista le falta a tu Claude ↓
17
9
72
Al final, el resultado tiene mucho más carácter que una presentación convencional del producto. A veces, cambiar la forma de contar una historia es justo lo que hace que una campaña destaque.
#capcutblackfriday
#capcutblackfriday
0
0
13
También ajusté los colores para mantener una estética más uniforme entre las escenas. Color Adjustment me ayudó a equilibrar la iluminación y conseguir que el video se sintiera como una sola pieza.
1
0
16
Después, organicé los clips en diferentes pistas con Multi-Track Editing. Así pude controlar mejor el orden de las tomas, ajustar su duración y hacer que cada detalle apareciera en el momento adecuado.
1
0
12
Primero, grabé algunos planos del producto y sus detalles, junto con elementos decorativos inspirados en el Día de Muertos. Quería que el producto siguiera siendo el protagonista, pero que el ambiente también contara una historia.
1
0
27
Black Friday también puede tener un toque de Día de Muertos. 💀🌼 Para esta campaña, quise transformar unas simples tomas de producto en un video con más personalidad usando CapCut PC.
tinyurl.com/2v23rn22
#capcutpc #capcutmarketing
tinyurl.com/2v23rn22
#capcutpc #capcutmarketing
5
3
188
OPUSCLIP 4.0 STOPPED BEING A CLIPPING TOOL AND HONESTLY IT WAS THE RIGHT MOVE
You drop one raw recording and it gives you back every version you need.
• The full edit for YouTube.
• A 60s clip for TikTok and Reels.
• A 3 to 5 min deep dive for X and LinkedIn.
• A quick .
The part I'd actually use is Recipes. You save your captions, cleanup and hook style once, and every new video comes out looking like the last one.
Recording was never the hard part. Doing the same edit every single week was.
They're giving free credits to everyone who engages with the original launch post.
You drop one raw recording and it gives you back every version you need.
• The full edit for YouTube.
• A 60s clip for TikTok and Reels.
• A 3 to 5 min deep dive for X and LinkedIn.
• A quick .
The part I'd actually use is Recipes. You save your captions, cleanup and hook style once, and every new video comes out looking like the last one.
Recording was never the hard part. Doing the same edit every single week was.
They're giving free credits to everyone who engages with the original launch post.
OpusClip 4.0 is here.
We used to find the best moments in your videos. Now OpusClip edits the whole thing, from the raw take to every format you post.
Turn on the edits you want, save your setup as a Recipe, and post everywhere. 🧵
opus.pro
4
1
89
🚨| UN LOCO LE HA QUITADO TODOS LOS ANUNCIOS A LA TELE DE SUS PADRES CON CLAUDE CODE Y AHORA PARECE UNA TELE NUEVA.
La tele estaba llena de anuncios, filas patrocinadas y apps que nadie había pedido.
Así que abrió Claude Code, le pasó el modelo de la tele y le pidió que la nectó a la tele por WiFi, encontró todas las apps basura y las fue desactivando una a una.
Después le montó una pantalla de inicio desde cero, negra, limpia y solo con las apps que usan.
Lo subió a X y se fue a más de 8 millones de visualizaciones.
Luego lo liberó todo gratis en GitHub y acabó siendo el repo número 1 en tendencias de la semana.
Funciona así: activas la depuración USB en los ajustes de la tele, clonas el repo y le pegas esto a Claude Code:
Inglés:
"I want to debloat my Android TV and replace its launcher with a clean, minimal home screen. My TV model is: [YOUR MODEL]. Work on the real device, step by step, and verify everything you do with screenshots pulled from the TV. Ask me before anything that depends on my preferences."
Español:
"Quiero limpiar mi Android TV de bloatware y cambiar su pantalla de inicio por una limpia y minimalista. El modelo de mi tele es: [TU MODELO]. Trabaja sobre la tele real, paso a paso, y comprueba todo lo que hagas con capturas sacadas de la propia tele. Pregúntame antes de cualquier cosa que dependa de mis preferencias."
En unos 15 minutos está hecho.
No borra nada, solo desactiva, y con una línea vuelves a dejar la tele como estaba.
Sin root y sin cables.
Ya hay más de 400 modelos probados de TCL, Sony, Xiaomi, Philips y más de 100 marcas.
Sirve para cualquier Android TV o Google TV, y si no tienes ordenador hay apps de la comunidad para hacerlo desde el móvil.
Te dejo el repo abajo 👇
La tele estaba llena de anuncios, filas patrocinadas y apps que nadie había pedido.
Así que abrió Claude Code, le pasó el modelo de la tele y le pidió que la nectó a la tele por WiFi, encontró todas las apps basura y las fue desactivando una a una.
Después le montó una pantalla de inicio desde cero, negra, limpia y solo con las apps que usan.
Lo subió a X y se fue a más de 8 millones de visualizaciones.
Luego lo liberó todo gratis en GitHub y acabó siendo el repo número 1 en tendencias de la semana.
Funciona así: activas la depuración USB en los ajustes de la tele, clonas el repo y le pegas esto a Claude Code:
Inglés:
"I want to debloat my Android TV and replace its launcher with a clean, minimal home screen. My TV model is: [YOUR MODEL]. Work on the real device, step by step, and verify everything you do with screenshots pulled from the TV. Ask me before anything that depends on my preferences."
Español:
"Quiero limpiar mi Android TV de bloatware y cambiar su pantalla de inicio por una limpia y minimalista. El modelo de mi tele es: [TU MODELO]. Trabaja sobre la tele real, paso a paso, y comprueba todo lo que hagas con capturas sacadas de la propia tele. Pregúntame antes de cualquier cosa que dependa de mis preferencias."
En unos 15 minutos está hecho.
No borra nada, solo desactiva, y con una línea vuelves a dejar la tele como estaba.
Sin root y sin cables.
Ya hay más de 400 modelos probados de TCL, Sony, Xiaomi, Philips y más de 100 marcas.
Sirve para cualquier Android TV o Google TV, y si no tienes ordenador hay apps de la comunidad para hacerlo desde el móvil.
Te dejo el repo abajo 👇
29
7
616
The #1 programming app on OpenRouter right now isn’t a general coding tool.
It’s Aura, an AI built only for game developers. And today they open-sourced Autopilot, the agent that plays your game and finds the bugs
It’s Aura, an AI built only for game developers. And today they open-sourced Autopilot, the agent that plays your game and finds the bugs
We're open-sourcing Autopilot.
A self-improving agent that plays your game and finds bugs. Unity, Godot and Unreal.
→ 43% more accurate
→ Almost 2X faster
→ Beats Claude Code at playtesting
It’s Computer Use for games – closing the agentic loop and solving the most tedious being your agent’s tester. Free trial + GitHub below.
4
4
113
🚨| KREA AD STUDIO IS THE FIRST AI TOOL THAT MAKES ADS YOU'D ACTUALLY RUN, AND I'M GOING TO SHOW YOU WHY.
Same hand soap. The first one is the AI slop we all scroll past. The second one came out of Krea Ad Studio.
Store Link → Products → Proven Layout → Real Brand Font → Ad r store link and it pulls your products on its own.
→ It builds every ad on a layout that already works.
→ The text is real type in your brand's font. No melted letters.
→ You edit it by talking. "Make it 9:16" and it's done.
→ It also works inside Claude and Grok.
Your first few ads are free.
Same hand soap. The first one is the AI slop we all scroll past. The second one came out of Krea Ad Studio.
Store Link → Products → Proven Layout → Real Brand Font → Ad r store link and it pulls your products on its own.
→ It builds every ad on a layout that already works.
→ The text is real type in your brand's font. No melted letters.
→ You edit it by talking. "Make it 9:16" and it's done.
→ It also works inside Claude and Grok.
Your first few ads are free.
introducing Ad Studio.
- powered by our new Krea 3 expert model for ads.
- paste a link of your brand and get dozens of ads
- full consistency with your brand and products.
try it now for free 👇
3
3
164
🚨| OPENROUTER JUST GOT A REAL ALTERNATIVE AND MOST TEAMS PAYING BIG INFERENCE BILLS HAVEN'T NOTICED YET.
RouterPlus doesn't route on list price.
It routes on your own evals and your real traffic.
Your Evals → Quality Bar → Best Model Per Task → Lowest Cost
• One or GPT, Claude, DeepSeek, Kimi, Qwen and GLM.
• You set the bar and every task goes to the cheapest model that clears it.
• Reserved capacity across providers on one contract, from 1 week to 12 months.
• 0% fee when you bring your own key. OpenRouter charges 5% past $25k/mo.
• 21× cheaper than the top single model on ITSMBench agent work.
204 models. 39 labs. 24 providers.
One key.
RouterPlus doesn't route on list price.
It routes on your own evals and your real traffic.
Your Evals → Quality Bar → Best Model Per Task → Lowest Cost
• One or GPT, Claude, DeepSeek, Kimi, Qwen and GLM.
• You set the bar and every task goes to the cheapest model that clears it.
• Reserved capacity across providers on one contract, from 1 week to 12 months.
• 0% fee when you bring your own key. OpenRouter charges 5% past $25k/mo.
• 21× cheaper than the top single model on ITSMBench agent work.
204 models. 39 labs. 24 providers.
One key.
AI teams overpay for inference twice: once for models bigger than they need, and again for idle capacity they don’t use. @useRouterPlus is the extended inference team for fast-growing AI companies. We test models and providers on your traffic and get providers to compete for nference better. Go to routerplus.com
2
4
113
video editors seeing "• 𝗣𝗿𝗶𝗰𝗶𝗻𝗴: Starting from $0.01/s at 720p"
Introducing P-Video-Upscale! 🎬
Fast video upscaling to 720p, 1080p, 2K, or 4K, while preserving the original aspect ratio and audio. Need smoother motion? Increase the frame rate to up to 60 fps.
And for launch, it's 50% off for the first two weeks, until November 12 at 18:00 eed to know:
• 𝗣𝗿𝗶𝗰𝗶𝗻𝗴: Starting from $0.01/s at 720p
• 𝗦𝗽𝗲𝗲𝗱: ~1.7s per second of output video, depending on resolution
• 𝗙𝗿𝗮𝗺𝗲 𝗿𝗮𝘁𝗲: Keep the original or interpolate up to 60 fps
• 𝗧𝘂𝗿𝗯𝗼 𝗺𝗼𝗱𝗲: Faster processing with a slight quality trade-off
No need to regenerate a video just to get it in higher resolution.
Available via our inference partners @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, Sociaro, and @runware.
Benchmarks by our benchmark partners @datapointai and @RapidataAI are coming soon.
📚 Model docs: docs.api.pruna.ai/guides/models/…
🧩 Try the model through the API or Playground: dashboard.pruna.ai/login
3
3
96
🚨| A 10 PERSON TEAM JUST BEAT OPENAI'S BEST SPEECH MODEL AND NOBODY SAW IT COMING.
Phonic-1 is out today and it's #1 on Artificial Analysis for speech-to-speech.
It's built for voice agents that actually get work done on a call.
Caller Talks • Agent Looks Up • Agent Acts • up info and takes actions while the caller is still talking.
• It switches between 50+ languages in the same voice.
• It holds up on noisy, real-world phone lines.
• It costs $0.06/min, cheaper than every leading alternative.
Building a voice agent? This one goes straight to the shortlist.
Phonic-1 is out today and it's #1 on Artificial Analysis for speech-to-speech.
It's built for voice agents that actually get work done on a call.
Caller Talks • Agent Looks Up • Agent Acts • up info and takes actions while the caller is still talking.
• It switches between 50+ languages in the same voice.
• It holds up on noisy, real-world phone lines.
• It costs $0.06/min, cheaper than every leading alternative.
Building a voice agent? This one goes straight to the shortlist.
Introducing Phonic-1, the most advanced speech-to-speech model for agentic tasks — from @phonic_co, the speech-to-speech model company.
We’ve had a breakout year and raised a $27.5M Series A led by @firstround with participation from @Lux_Capital, @felicis, and Definition.
4
3
137
🚨| SUNRA JUST KILLED THE AI MARKUP.
All the top models in one place, at the same price the labs themselves charge.
One API Key → Every Model → Creator's Price → Your App.
→ Image, video, audio and 3D models with a single API key.
→ Every model at its creator's API ay only for what you use, no subscription.
→ 10,000 Seedance 2.5 clips cost $11,652 on Sunra vs $23,112 on fal.
→ Working at scale? The Enterprise plan gets you up to 40% off the API price of the top models.
Here's a video I made with their API, it speaks for itself… 👇
All the top models in one place, at the same price the labs themselves charge.
One API Key → Every Model → Creator's Price → Your App.
→ Image, video, audio and 3D models with a single API key.
→ Every model at its creator's API ay only for what you use, no subscription.
→ 10,000 Seedance 2.5 clips cost $11,652 on Sunra vs $23,112 on fal.
→ Working at scale? The Enterprise plan gets you up to 40% off the API price of the top models.
Here's a video I made with their API, it speaks for itself… 👇
Today we're launching the Sunra API.
Every leading image, video, audio and 3D model through one key, each at its maker's own API price. Pay as you go, with no subscription and no markup.
The models are ready. Your app is next.
8
4
45
wait... ai automates comments now??
We gave you distribution, now we’re giving you conversion! 🙇
Manage DMs and comments across your fleet of organic accounts in 15+ countries.
ManyChat-style automation, included with every TokPortal account at no extra cost:
→ Reply to comments
→ Build automated reply and Turn viewers into customers with DM automation
Enjoy our first TokPortal Direct! 🙇
4
2
124
higgsfield since august:
> 30M users → 40M+
> crosses $1B annualized
> 150M generations a month
> nvidia and tcv (netflix, spotify) come in
not bad for a video app
> 30M users → 40M+
> crosses $1B annualized
> 150M generations a month
> nvidia and tcv (netflix, spotify) come in
not bad for a video app
@JensenHuang and NVIDIA just invested in Higgsfield
We turn competition at the foundation layer into compounding advantage for our customers. Every advance at the frontier expands what Higgsfield can deliver
The moat is ownership of demand, and we clearly own one.
2
1
138
average video editor's reaction reading this tweet
Introducing P-Video-2-Pro-Avatar, based on @MiniMax_AI H3 !
Turn a single portrait into a realistic talking video with natural expressions and accurate lip sync.
And for launch, it's 50% off for the first two weeks, until November 5 at 18:00 CET!
• 𝗣𝗿𝗶𝗰𝗶𝗻𝗴: $0.02/s 8p
• 𝗦𝗽𝗲𝗲𝗱: ~0.97 s at 480p · ~2.2 s at 768p per second of output video
• 𝗜𝗻𝗽𝘂𝘁: One portrait with a text script or uploaded audio
• 𝗩𝗼𝗶𝗰𝗲𝘀: 30 voices across 10 languages
• 𝗗𝘂𝗿𝗮𝘁𝗶𝗼𝗻: 1s–3min at 24 fps, in 480p or 768p
• 𝗖𝗼𝗻𝘁𝗿𝗼𝗹: Voice delivery, scene motion, seed, and prompt upsampling
One image, two ways to bring it to life. Use your own audio or generate speech directly from text.
Available via our inference partners @replicate, @eachlabs, @Cloudflare, @scenario_gg, @wavespeed_ai, @WiroAI, Sociaro, @TellersAI, and @togethercompute.
📚 Model docs: docs.api.pruna.ai/guides/models/…
🧩 Try the model through the API or Playground: dashboard.pruna.ai/login
0
1
15