ney's back retweeted
MY FEED RUNS ON CLAUDE.
My clients run on PowerPoint.
Two different planets.
→ Agent inside PowerPoint
→ Knows your brand
→ Shared with your team
→ 2 hours to 8 minutes
Templafy Document Agents just connected them.
🤝 Paid partnership
AI that knows your company’s content and branding? 👀
With Templafy Team, build and share AI Agents that create editable PowerPoints using your company's knowledge and templates.
Get started 👉 templafy.com/templafy-teams-…
ney's back retweeted
CODING AGENTS AREN'T SLOW. REVIEW IS.
agents already write faster than you can read
axiom's swarm on cfsg:
→ 108 milestones, ~380 tasks
→ claimed via GitHub issues
→ every result checked by lean
→ 43.1M lines of code
code a checker signs off on scales. code you have to read doesn't
that's axiom's bet
🤝 Paid partnership
ney's back retweeted
UNA STARTUP QUIERE MATAR EL MODERN DATA STACK
durante años la lógica de tus datos ha vivido dentro del producto de otro
difícil de mover y difícil de entender para una IA
querio acaba de lanzar con $5M de seed:
→ tu lógica como código en tu GitHub
→ un repo open source para migrar tu BI
→ mcp y cli para conectar agentes
la idea es que los datos y su lógica sean tuyos, y que las herramientas se enchufen
🤝 Paid partnership
ney's back retweeted
if the agent says it sent the email, i would also enjoy the email having been sent
Today, @JudgmentLabs opens for general access.
We believe long running agent systems will generate more exhaust data than the world can currently process.
Hidden within agent traces are rich records of how to make these systems better: what users ask for, when agents stumble, and how agents interact with the external world.
Judgment is the most performant monitoring and improvement layer for long-horizon agents.
We’re excited to see what you build!
ney's back retweeted
YOUR AI CAN NOW WORK ON A PLANE WITH NO WIFI
Nativ, by Neywa Labs, is a free Mac app:
→ Download a model
→ Turn off the internet
→ It keeps working
→ Nothing leaves your Mac
No subscriptions, no one else's servers.
🤝 Paid partnership
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
ney's back retweeted
OpenAI's CEO realizing one guy built a library with 5M+ downloads that runs AI on a Mac without touching a single API
🤝 Paid partnership
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
ney's back retweeted
We pay to run AI in datacenters while the Mac on our desk can already run many of those models.
Neywa Labs comes out of stealth today to use that power: local, open AI on your own machine.
The datacenter, on your desk.
🤝 Paid partnership
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
ney's back retweeted
your dad is in his 60s. a cardiologist gives him a stress test
the results come back bad. serious heart disease that needs immediate treatment
nobody calls him
a few days later the pain starts in his neck and shoulder. he ignores it. why wouldn’t he? as far as he knows, his heart is fine
8 days after the test he’s in the ER. heart attack. three major arteries almost completely blocked
he dies the next day
this isn’t a hypothetical. it happened to a man in south carolina
the clinic didn’t call because, in its view, he was responsible for contacting them. his family sued and a jury sided with them
their expert said if he’d been told, he most likely would have recognized the symptoms and lived
the test caught it. the result just never left the building
> in one analysis of 23,000+ malpractice cases, communication failures were a factor in 30% of them, including 1,744 deaths
this is what AI in healthcare should be pointed at. not replacing doctors. making sure what the hospital finds actually reaches the patient
🤝 Paid partnership
ney's back retweeted
🚨THIS JUST KILLED MISSED DIAGNOSES
10M+ patients are already on it and almost no one is talking about it.
It's called Concurrence, and it's the AI that lets every hospital learn from every patient.
It allows hospitals to:
→ Call patients in their own language
→ Catch the problems the chart misses
→ Run millions of care scenarios before the doctor picks a plan
→ Handle prescriptions, instructions, and 30-day check-ins
→ Get smarter with every single outcome
No patient falls through the cracks.
No patient is the first case their doctor has seen.
All inside the hospital's own system.
🤝 Paid partnership
ney's back retweeted
New models ship every week, but few are ready to run on a laptop on launch day.
Neywa Labs works with model labs to bring them to Apple silicon and MLX from day one.
They've already done it for models from 8 labs.
🤝 Paid partnership
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
ney's back retweeted
Most AI runs on someone else's servers, with per-token costs and your data off your device.
Neywa Labs' omni engine lets you build apps that read, see, hear, speak and create on your hardware.
7M+ downloads behind it.
🤝 Paid partnership
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
ney's back retweeted
just saw this and i'm still shook.
the creator of mlx-vlm built a free Mac app that runs 5,000+ models locally.
ask it to change something in a doc and it edits it right there, no cloud.
download the model once and it keeps working with the wifi off.
it's called Nativ.
🤝 Paid partnership
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
ney's back retweeted
YOU CAN NOW RUN YOUR OWN AI ON YOUR MAC, FREE AND OFFLINE
It's called Nativ, by Neywa Labs:
→ 5,000+ models
→ Free and open source
→ Your data stays local
→ No monthly fees
Your AI, on your computer.
🤝 Paid partnership
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
ney's back retweeted
COMPLETE 31-MIN GUIDE TO BUILDING A TRADING BOT WITH CLAUDE FABLE 5.1
Best trading bot tutorial I've seen in 2026.
Claude → Bot → Strategy → Live
🔖 Save it before you lose it.
🚨 This is SERIOUSLY dangerous.
I just watched a 19-year-old kid casually explain how he makes $10,000 a month with AI while he sleeps, and I'm convinced that 99% of people still don't get this.
The whole system:
Pick a YouTube channel that uploads content constantly.
Drop the link into a tool.
Connect your TikTok, Instagram, and YouTube Shorts accounts.
Shut the laptop.
From there, the AI handles everything.
It detects every new video that goes live, finds and cuts the moments with the most viral potential, adds captions, and posts them automatically across every platform.
1 million views ≈ $2,000
Zero editing.
Zero manual uploads.
Zero stress.
10 minutes to set up and then it runs on full autopilot.
You're not creating content anymore.
You're building a content machine.
Bookmark this.
ney's back retweeted
LIQUID AI JUST DROPPED 6 MODELS THAT RUN ON YOUR LAPTOP
LFM3, all ~2B parameters:
→ Text for agents
→ Vision with video
→ Voice in 18 languages
→ Omni: sees, hears, talks
→ One-pass decisions
In their tests, the 1.7B one beats Qwen3.5-9B on WebArena-Verified.
🤝 Paid partnership
Introducing 𝙻𝙵𝙼𝟹: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.
> 6 model instances, across text, audio, vision, and omni modalities, generative and decision models.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱: text, 256k context window.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱: vision-language model with video-input support for spatial understanding
> 𝙻𝙵𝙼𝟹-𝙰𝚞𝚍𝚒𝚘-𝟸𝙱: tool calls, voice chat, reasoning in 18 languages.
> 𝙻𝙵𝙼𝟹-𝙾𝚖𝚗𝚒-𝟸𝙱: an omni model combines vision, audio, and language.
> 𝚍𝟷-𝟸𝙱 and 𝚍𝟷-𝙾𝚖𝚗𝚒-𝟸𝙱: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> 𝙻𝙵𝙼𝟹-𝚅𝙻-𝟸𝙱 is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> 𝙻𝙵𝙼𝟹-𝟷.𝟽𝙱 is #1 on the Pareto frontier of device intelligence 𝚙𝚒𝚙𝚎𝚝𝚝𝚎.𝚕𝚒𝚚𝚞𝚒𝚍.𝚊𝚒
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ 🧵
ney's back retweeted
A GENERAL AGENT WITH ONE INSTRUCTION JUST BEAT THE EXPERT KERNEL AGENTS
no handcrafted search workflow. no kernel-specific knowledge base. no stack of specialized sub-agents.
it's called JustGoal.
gpu kernels written on nvidia b200 by a general agent given only an optimization objective
the agent builds its own evaluation loop and iterates
paper + full operator code public, apache 2.0
outperforms open-source libraries on 8 of 10 operators, per-operator geomean speedup 1.14–2.34×.
→ b200, sm_100a
→ beats the fastest specialized-agent kernel on 5 of 9 operators
→ in same-model comparison, the kda-0.5 harness lands at 0.95× justgoal
→ ablations + all code shipped with the paper
yesterday's answer, retired by one prompt.
🤝 Paid partnership
Do we still need specialized agents for GPU kernel optimization?
We asked the question, and ran the experiment to answer it.
New paper with @THU_NLP: Agents under Goal mode are now able to beat systems deliberately built for kernel optimization. We named it JustGoal.
On FlashAttention, Gated DeltaNet, and DeepSeek Sparse Attention, a general frontier model (in our case, using Fable 5.1) delivers 1.08× to 1.58× speedup over expert-written references. Against the specialized KDA-0.5 (1.45× and 1.28×), Fable 5.1 hits 1.58× on both.
Frontier models are really burying the talent of kernel optimization experts in the past.
Paper: github.com/OpenBMB/JustGoal-…
ney's back retweeted
WAIT. we also need to check the thing checking the agent? 😭
🤝 Paid partnership
Today, @JudgmentLabs opens for general access.
We believe long running agent systems will generate more exhaust data than the world can currently process.
Hidden within agent traces are rich records of how to make these systems better: what users ask for, when agents stumble, and how agents interact with the external world.
Judgment is the most performant monitoring and improvement layer for long-horizon agents.
We’re excited to see what you build!
ney's back retweeted
POWEPOINT ALREADY SPEAKS YOUR BRAND'S LANGUAGE
the one who sells, the one who advises, the one who closes the numbers. everyone ends up building decks by hand.
Templafy Document Agents start from your content and your brand. not from a generic deck built from scratch.
→ they work inside PowerPoint
→ they create, edit and review
→ according to Templafy, from 2 hours to 8 minutes
ai doesn't replace PowerPoint.
it improves it where you already work.
according to Templafy, more than 4 million enterprise users already build their documents with them.
🤝 Paid partnership
AI that knows your company’s content and branding? 👀
With Templafy Team, build and share AI Agents that create editable PowerPoints using your company's knowledge and templates.
Get started 👉 templafy.com/templafy-teams-…
ney's back retweeted
CADA DEMO TUYO PUEDE VERSE COMO UN ANUNCIO DE $5.000 Y NO CUESTA NADA
La mayoría de founders suben demos con la pantalla tal cual, sin zoom y con el cursor saltando por todos lados.
Y la gente se va a los 10 segundos.
OpenScreen arregla eso solo.
Le das a grabar, haces tu demo normal y la herramienta se encarga del resto:
→ Hace zoom justo donde haces clic
→ Suaviza el movimiento del cursor
→ Añade motion blur en cada transición
→ Pone tu cara en una burbuja de webcam
→ Mete un fondo bonito con sombras
→ Te deja anotar con texto y flechas
→ Exporta en el formato que te pida cada red
Lo usas para lanzamientos, landing pages, tutoriales, outreach en frío o el vídeo de bienvenida de tu producto.
Lo que antes era Screen Studio por $29 al mes ahora es un repo con 39.9K estrellas.
Licencia MIT, así que puedes usarlo en proyectos comerciales.
Repo abajo
Guárdalo antes de tu próximo lanzamiento 🔖
ney's back retweeted
YOU CAN NOW EDIT VIDEO WITHOUT LEAVING THE CHAT IN CLAUDE AND CHATGPT
the plugin is called Cardboard and just launched.
drop in your clips or a reference video, describe the edit you want and it builds it right there in the conversation.
→ captions
→ music
→ motion graphics
→ voiceover
→ Recipes for social, launches, trip montages
if something feels off, you tell it in the chat and it reworks it.
what changes here isn't "making video with AI". it's that the chat where you thought up the idea now ships it too.
link below.
🤝 Paid partnership
Introducing Cardboard for ChatGPT and Claude.
You never know when your best ideas will find you.
So we made sure that @usecardboard is there, whenever they do.
At your desk, or on the go, wherever you are :)