DI.GG
RankingsTrendingAlpha
DI.GG/AI
Feed
trendingtopnewbookmarked
Rankings
AI 1000 (Humans)CompaniesPoliticians
Slices of Data
ResearchersCombativeDevil's AdvocateSupportiveTeaching
DI.GG/AI
Feed
trendingtopnewbookmarked
Rankings
AI 1000 (Humans)CompaniesPoliticians
Slices of Data
ResearchersCombativeDevil's AdvocateSupportiveTeaching
DI.GG/AI

Elon Musk Shares Reluctant Journey to Tesla CEO Role in 2008

23h
TLDR
Elon Musk reposts an account detailing his early reluctance to become Tesla's CEO, emphasizing his identity as an engineer and product leader who realized full product control required company leadership. Initially serving as chairman and funder while Martin Eberhard was CEO from 2002 to 2008, Musk stepped in during Tesla's near-collapse in 2008 despite also leading SpaceX. The narrative highlights the extraordinary challenge of simultaneously guiding two high-stakes tech ventures to success.
Engaged Users
OP
EM
Elon Musk@elonmusk
#3 AICEO
54.4M3.2k7.7k9.4k

Amazon Caught Manipulating Prices via Brands and Competitors

14h
→ @Ric_RTP
TLDR
Court documents in a California lawsuit reveal Amazon monitored competitors' prices in real time and pressured brands like Levi's, Hanes, and Allergan to raise those prices elsewhere. Brands then contacted retailers such as Walmart, Target, and Home Depot to increase prices, allowing Amazon to match higher rates and suppress lower-priced listings. This scheme allegedly spanned years and categories, leading to higher costs for consumers amid allegations of illegal price fixing.
Engaged Users
OP
RI
Ricardo@Ric_RTP
RT
PG
Paul Graham@paulg
#18 AIMedia
1.5M1.3k19k8.9k

Stitch by Google Open-Sources DESIGN.md Specification for Design Rules

11h
TLDR
Stitch by Google open-sources the draft specification for DESIGN.md, enabling export and import of design rules like colors and typography across tools and platforms. The format allows agents to understand design intents and validate choices against WCAG accessibility rules. David East demonstrates its use in VS Code, including YAML frontmatter for tokens and GitHub integration.
Engaged Users
OP
SB
Stitch by Google@stitchbygoogle
#705 AI
RT
KM
Kory Mathewson@korymath
#643 AIResearcher
QT
JW
Josh Woodward@joshwoodward
#473 AIExec

“DESIGN.md is one of my favorite things that Stitch is doing! If you know, you know.

848.9k895727.9k

Elon Musk Endorses Philosophy: Best Products Need No Logo

13h
TLDR
Marc Andreessen reveals Elon Musk's philosophy that the world's best products shouldn't need logos, emphasizing relentless quality as in top cars. Musk responds with Shakespeare's 'A rose by any other name would smell as sweet,' affirming that product essence trumps branding. The exchange highlights Musk's product-first approach in tech ventures like Tesla.
Engaged Users
OP
EM
Elon Musk@elonmusk
#3 AICEO
18.6M2.3k5.5k5k

Hugging Face Launches ml-intern, Open-Source Agent Automating ML Research Loop

17h
→ @akseljoonas
TLDR
Hugging Face introduces ml-intern, an open-source AI agent that automates the ML post-training research process by analyzing papers, handling datasets, implementing ideas, and training models via the Hugging Face ecosystem. It excels in tasks like scientific reasoning on GPQA (32% score vs. Claude Code's 22.99%), healthcare data generation outperforming Codex, and competitive math training with GRPO. Released as a CLI tool and web app, it enables autonomous, research-backed model development on阿
Engaged Users
OP
AK
Aksel@akseljoonas
RT
C🤗
clem 🤗@ClementDelangue
#111 AICEO
QT
EL
elvis@omarsar0
#833 AIBuilder

“Karpathy's autoresearch repo started an impressive trend. Agents can now train AI models to build SoTA agentic systems. And to think this is just scratching the surface. Ultimately, it boils down to good research questions or hypotheses. LLMs are not great at this (yet).

QT
TW
Thomas Wolf@Thom_Wolf
#62 AIFounder

“Love this work from Aksel and the post-training team at Hugging Face! Turns out the HF ecosystem (papers, datasets, models all accessible through CLI, skills and md files) is perfect for running SOTA ML agents: agents that can train any type of AI model to top performance. A few concrete runs: ⭐️ Scientific reasoning: the agent walked citations from the benchmark paper, pulled OpenScience and NemoTron-CrossThink, added 7 difficulty-filtered variants from ARC/SciQ/MMLU, and ran 12 SFT ablations on Qwen3-1.7B. GPQA went from 10% to 32% in under 10 hours. Claude Code's best on the same prompt was 22.99%. ⭐️ HealthBench: it judged the existing datasets too noisy (!), generated 1100 synthetic examples covering emergencies, hedging and multilingual cases, upsampled 50x, and beat Codex by 60% (careful to check overfitting here) ⭐️ Competitive math: wrote a full GRPO script, launched A100s on HF Spaces, watched rewards climb and then collapse, and ran ablations until it found a recipe that held. And the harness is pretty tiny and simple. A couple of best practices and a handful of skills pointing at tools already in the ecosystem: arxiv and http://hf.co/papers for reading, the Hub for datasets and models, HF Jobs for compute, Trackio for metrics. Personal favorite is the "research skill" explaining how to do a SOTA landscape of a field (see https://github.com/huggingface/ml-intern/blob/main/agent/tools/research_tool.py) which is extremely powerful when combined with a simple prompt that basically tell "FIRST: Search HF ecosystem to find the best approach) (see https://github.com/huggingface/ml-intern/blob/main/agent/prompts/system_prompt.yaml#L14) On another note: setting good baselines on new benchmarks keeps getting harder when a setup this simple beats raw Codex by 60% on HealthBench out of the box. Give it a try if you're training AI models. We provisioned $1k of GPU resources and Anthropic credits for the quickest among you. Links: Github (CLI): https://github.com/huggingface/ml-internWeb Spaces (mobile): https://huggingface.co/spaces/smolagents/ml-intern

QT
LT
Lewis Tunstall@_lewtun

“OpenAI's first AI intern is expected by the end of this year, but we got impatient and decided to build it ourselves :) > Runs autonomously for hours / days depending on the task. > Can read every paper, model, and dataset on the HF Hub to build the best post-training recipes > Works with any capable model (Kimi K2.6, GPT-5.4, Opus 4.7 etc) > Runs locally or on HF infra (Spaces as sandboxes, Jobs for generating data & training models, Buckets for storage)

QT
MP
Mayank Pratap Singh@Mayank_022

“I tested @huggingface ml-intern, given the prompt "Fine-tune a Segment Anything Model (SAM) on a useful medical dataset. Train the model, and provide a comprehensive tutorial in a Jupyter Notebook file. Additionally, create a Hugging Face article/blog post documenting everything you have done." It did it all autonomously: - Researched via hf_papers & searched GitHub/HF Hub - Found an HF dataset & wrote the finetuning script - Trained it using HF compute (took ~1 hour) - Pushed the weights & wrote the article Here are the model weights, code, and the blog it generated: hf article https://huggingface.co/Mayank022/blog-fine-tuning-sam-medical-segmentation model weights https://huggingface.co/Mayank022/sam-vit-base-kvasir-polyp-segmentation Awesome stuff @akseljoonas , looking forward to use this. 🔥

QT
CM
Carlos Miguel Patiño@cmpatino_

“I’ve been using ml-intern for a while, and it genuinely changed my workflow. It's super good at: - Model/Dataset discovery. - Post-Training setup iteration. - Data processing workflows. Huge shoutout to @akseljoonas for leading this!

376.2k963973.9k

EU Court Rules Hungary Violated Law Banning Children from LGBTQ+ Content

12h
→ @durov
TLDR
The Court of Justice of the EU has ruled that Hungary violated EU law by enacting a ban preventing children from accessing LGBTQ+ content. Telegram CEO Pavel Durov criticizes the decision sarcastically, highlighting what he sees as inconsistency in EU child protection policies that restrict teens from social media but allow such content. This ruling emphasizes ongoing tensions between national safeguards for minors and broader EU free speech and single market rules.
Engaged Users
OP
PD
Pavel Durov@durov
CEO
RT
EM
Elon Musk@elonmusk
#3 AICEO
1.8M1.1k11.6k3.1k

OpenAI Launches ChatGPT Images 2.0 for Complex Visual Tasks

8h
TLDR
OpenAI introduces ChatGPT Images 2.0, a state-of-the-art image generation model designed for complex visual tasks, delivering precise visuals with sharper editing, richer layouts, and advanced intelligence. The model produces immediately usable outputs, as showcased in a 60-second video montage featuring creative scenes like a chameleon-themed story generated entirely with the tool. This upgrade enhances ChatGPT's multimodal capabilities, targeting creators and developers needing high-fidelity,智
Engaged Users
OP
OP
OpenAI@OpenAI
#1 AICompany
RT
PD
Prafulla Dhariwal@prafdhar
#152 AIResearcher
QT
GB
Greg Brockman@gdb
#9 AIExec

“really incredible what you're now able to create with a little bit of compute. excited for new applications in areas like education, professional settings (slides, marketing materials, etc), and productivity (such as creating diagrams for code documentation).

QT
NR
Nous Research@NousResearch
#87 AICompany

“Access GPT Image 2.0 natively in Hermes Agent Update now to get access - just run `hermes update` and select your image generation tool model with `hermes tools`

QT
YJ
Yuchen Jin@Yuchenj_UW
#805 AIBuilder

“Just tried gpt-image-2. It is really good. OpenAI is finally leading the image gen again.

QT
AC
Andrew Curran@AndrewCurran_
#793 AIMedia
QT
TH
Tim Hwang@timhwang
#690 AIPolicy

“this is going to absolutely revolutionize the garfield intelligence layer

QT
SH
shyamal@shyamalanadkat
#920 AIMedia

“broke the internet again

1.8M6291.6k3k

xAI Partners with Cursor to Build Leading Coding AI Using Colossus Supercomputer

6h
TLDR
xAI and Cursor AI announce a close collaboration to develop the world's best AI for coding and knowledge work. The partnership combines Cursor's top product and distribution to software engineers with xAI's Colossus supercomputer, equivalent to a million H100 GPUs. Cursor grants xAI an option to acquire it later this year for $60 billion or pay $10 billion for their joint efforts.
Engaged Users
OP
SP
SpaceX@SpaceX
#22 AICompany
RT
SO
Soleio@soleio
Media
QT
@J
@jason@Jason
#500 AIMedia

“🔥 wow! Very strategic and bold move colossus is a super weapon for SpaceX already — can you imagine when it scale to the the stars?!?

QT
AA
Austen Allred@Austen
#696 AICEO

“OK this could actually work. Makes much more sense than Cursor trying to train its own models in house.

QT
B(
Beff (e/acc)@beffjezos
#285 AIMedia

“"Ok Cursor, tiles the lightcone with Optimus bots, make no mistakes"

QT
S🇸🇬
swyx 🇸🇬@swyx
#251 AIMedia

““Cursor has also given SpaceX the right to acquire Cursor later this year for $60 billion or pay $10 billion for our work together.” personally this is the most exciting option pricing deal of the year, wow, kudos to both sides!!

QT
DS
Dan Shipper 📧@danshipper
CEO

“huge huge deal

QT
MT
Michael Truell@mntruell
#312 AICEO

“Excited to partner with the SpaceX team to scale up Composer. A meaningful step on our path to build the best place to code with AI.

4.9M1.6k3.1k2.9k

Bud Launches AI Human Emulator with Autonomous Computer Capabilities

13h
→ @budapp
TLDR
Bud introduces an AI agent called Bud, described as the first AI Human Emulator, equipped with its own full computer environment including storage, compute, memory, browser, and communication tools like SMS and Telegram. It can autonomously build and code applications, process files, research and generate reports, create custom apps from PDFs, and handle tasks end-to-end like a human. Users can access it via text or at bud.app with a free trial, and commenting offers 100k free credits.
Engaged Users
OP
BU
Bud@budapp
RT
RS
Robert Scoble@Scobleizer
#435 AIMedia
392.9k2.1k2262.8k

Linus Torvalds Calls Elon Musk's Government Efficiency Idea 'Too Stupid'

14h
→ @realBigBrainAI
TLDR
A video clip features Linux creator Linus Torvalds dismissing a recent U.S. government efficiency initiative as 'too stupid' during an interview. The interviewer points out the idea is linked to Elon Musk, who co-leads the Department of Government Efficiency (DOGE). The post frames Torvalds' comment as unintentionally criticizing Musk, sparking viral attention among tech enthusiasts.
Engaged Users
OP
BB
Big Brain AI@realBigBrainAI
RT
YL
Yann LeCun@ylecun
#6 AIResearcher
725.5k1412.5k2.7k

X Launches Custom Timelines Powered by Grok for Premium Users

4h
TLDR
X introduces Custom Timelines, enabling users to pin specific topics to their home tab with support for over 75 niches. The feature leverages Grok's analysis of posts combined with algorithmic personalization for tailored feeds, performing best on engaged topics. Early access rolls out to Premium subscribers on iOS, with Android support imminent.
Engaged Users
OPQT
NB
Nikita Bier@nikitabier
#176 AITechnologist
QT
RS
Robert Scoble@Scobleizer
#435 AIMedia

“Looking through the new AI feed on my iPhone and first impressions are that it is very good. Very little noise. Good diversity of what it is showing. Will help new people see what is going on in AI, which is really good. Lists still have their role, especially in doing things with the X API and my AI news site, https://alignednews.com/ai which is wholly made from X lists, is a better way for someone who only has 10 minutes a day to keep up on all the important news. I haven’t checked out the others yet, just the AI one, but it is really good. So I am happy. What do you think?

RE
B(
Beff (e/acc)@beffjezos
#285 AIMedia

“@nikitabier @Scobleizer You mad lads actually built it 🥹

1.1M2.9k1.6k2.5k

Brex Launches CrabTrap, LLM Proxy Securing AI Agents in Production

13h
→ @pedroh96
TLDR
Brex announces CrabTrap, an open-source proxy that uses LLMs to intercept outbound requests from AI agents and block risky activity before reaching external APIs. Developed to address the lack of safe production deployments for fast-growing agent projects like OpenClaw, it enables secure internal use without granting real credentials. Early results show promise for enhancing agent harness security at scale, with Brex now hiring for similar hands-on engineering roles.
Engaged Users
OP
PF
Pedro Franceschi@pedroh96
RT
PS
Peter Steinberger 🦞@steipete
#300 AIBuilder
QT
GT
Garry Tan@garrytan
#130 AICEO

“CrabTrap is a big deal for the OpenClaw community

QT
JF
Jared Friedman@snowmaker
VC

“Brex just open sourced the key piece of infrastructure that enabled them to run their whole company on OpenClaw.

455.8k691032.4k

Indian Tanker Pays Fake IRGC Crypto Scammers, Faces Real Attacks in Strait of Hormuz

17h
→ @FatManTerra
TLDR
An Indian oil tanker, Sanmar Herald, reportedly fell victim to a cryptocurrency scam impersonating Iran's IRGC, paying USDT for safe passage through the Strait of Hormuz. Instead, the vessel encountered gunfire from actual IRGC forces, pleading for them to stop via audio recordings. This incident marks a potential first where a crypto transaction directly preceded maritime conflict.
Engaged Users
OP
FA
FatMan@FatManTerra
RT
DF
Danielle Fong 🔆@DanielleFong
#609 AIFounder
QT
T(
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex
#586 AIMedia

“So those Indians got scammed. Only pay in crypto to certified Islamic Revolution Guard Corps, folks! (Funny thing: if the IRGC publishes their official USDT wallets, I suspect they'll get sanctioned)

1.4M2852.1k2.4k

OpenAI Developers Introduce Euphony, Open-Source Tool for Visualizing Chat Data

12h
TLDR
OpenAI Developers released Euphony, an open-source web tool that visualizes chat data and Codex session logs from public URLs or local files. It offers interactive features like translation, filtering, editing, tokenization views, and session file trees, supporting Hugging Face datasets and models such as o200k_harmony and gpt-5.4. The tool aids developers and researchers in analyzing AI conversations, metadata, and code sessions with rendered markdown and diagrams.
Engaged Users
OP
OD
OpenAI Developers@OpenAIDevs
#79 AICompany
RT
AD
Adam.GPT@TheRealAdamG
#489 AIExec
QT
SW
Simon Willison@simonw
#194 AIBuilder

“OpenAI's new Euphony tool works almost exactly the same way as my Codex transcript viewer https://tools.simonwillison.net/codex-timeline?url=https%3A%2F%2Fgist.githubusercontent.com%2Fsimonw%2Fa9eb5993a2853ec840d26c0e56bde362%2Fraw%2Fb8c5febdf60d878da84e27c07efdaed159abde4a%2Flogs.jsonl#tz=local&q=&type=all&payload=all&role=all&hide=1&truncate=1

QT
BB
Boaz Barak@boazbaraktcs
#139 AIResearcher

“I have found Euphony so useful internally! Glad it's now open source!

274.2k802682.4k

Google Open-Sources Demo for Running Multiple Gemma 4 Instances Locally

12h
TLDR
Google's Gemma team releases an open-source demo enabling users to run 3, 5, or 10 concurrent instances of Gemma 4 models on local hardware. The demo showcases Gemma 4 26B A4B handling 10+ concurrent requests on a MacBook Pro M4 Max at 18 tokens per second per request. A video demonstrates multiple agents generating SVG images simultaneously, highlighting impressive local inference performance.
Engaged Users
OP
GG
Google Gemma@googlegemma
#678 AI
RT
OS
Omar Sanseviero@osanseviero
#903 AITechnologist
RT
CF
Clément Farabet@clmt
#266 AIExec
QT
DF
Danielle Fong 🔆@DanielleFong
#609 AIFounder

“talk talk talk

522.5k652802.4k

OpenAI Demonstrates ChatGPT Images 2.0 in Showcasing Video

9h
TLDR
OpenAI shares a demo video highlighting the capabilities of ChatGPT Images 2.0, featuring a diverse array of AI-generated visuals. The sequence includes a museum curator scene, a global internet adoption chart, an encyclopedia page with da Vinci sketches, product packaging, a manga comic, an optical illusion with rice grains, a child watching Apollo footage on vintage TV, and the ChatGPT logo. This update emphasizes advanced image generation for realistic, stylized, and data-driven content.
Engaged Users
OP
OP
OpenAI@OpenAI
#1 AICompany
RT
RS
Robert Scoble@Scobleizer
#435 AIMedia
RT
JK
Jason Kwon@jasonkwon
#90 AIExec
QT
RP
Ryan Petersen@typesfast
#633 AICEO

“Shutterstock share price up 12% in the last year...

QT
DU
Derya Unutmaz, MD@DeryaTR_
#781 AIMedia

“It’s on !

QT
AR
Arena.ai@arena
#13 AICompany

“Exciting news - GPT-Image-2 by @OpenAI has claimed the #1 spot across all Image Arena leaderboards! A clean sweep with a record-breaking +242 point lead in Text-to-Image - the largest gap we’ve seen to date. - #1 Text-to-Image (1512), +242 over #2 (Nano-banana-2 with web-search aka gemini-3.1-flash-image) - #1 Single-Image Edit (1513), +125 over #2 (Nano-banana-pro aka gemini-3-pro-image) - #1 Multi-Image Edit (1464), +90 over #2 (Nano-banana-2) No model has dominated Image Arena with margins this wide. Huge congratulations to @OpenAI on this major breakthrough in image generation! More performance breakdowns by category in the thread below.

QT
DF
Daniel Fradin@danielfradin

“made this film for the images 2.0 launch. entirely made with generated images!! new model is insane @gabeeegoooh and team went crazy with this one

2.2M5329352.3k

Apple CEO Tim Cook Introduces John as Successor

8h
TLDR
Apple CEO Tim Cook expresses gratitude to supporters for his leadership and introduces John as his successor in a heartfelt message. He emphasizes that the company remains centered on users and shares a photo of the two walking together at Apple Park. The announcement signals a major leadership transition at the tech giant.
Engaged Users
OP
TC
Tim Cook@tim_cook
#270 AICEO
QT
SP
Sundar Pichai@sundarpichai
#36 AICEO

“Congrats on an incredible run @tim_cook , always respected your deep commitment to Apple's mission and best wishes in your new role! Look forward to working with John as well!

RE
SN
Satya Nadella@satyanadella
#63 AICEO

“@tim_cook Wishing you all the best in this next chapter, @tim_cook. You’ve always led with clarity and purpose, and your impact on Apple and across our industry will be felt for years to come. And congrats John!

1.7M2.3k5.2k2k

Odyssey Unveils Odyssey-2 Max, Most Powerful World Model Advancing Physical Accuracy

9h
TLDR
Odyssey launches Odyssey-2 Max, its most advanced world model, which significantly surpasses state-of-the-art benchmarks in physical accuracy and realism. The model supports real-time simulation and interaction with the physical world, moving beyond traditional language models toward comprehensive intelligence. It positions as foundational technology for applications including defense, demonstrated through interactive reality simulations.
Engaged Users
OP
OD
Odyssey@odysseyml
#477 AI
RT
RS
Robert Scoble@Scobleizer
#435 AIMedia
RT
SO
Soleio@soleio
Media
QT
NB
Nathan Benaich@nathanbenaich
#390 AIVC

“big world model tings!

1.1M1491941.9k

Tesla Coil Ignites Plasma Inside Syringe via Paschen’s Law

20h
→ @mathemetica
TLDR
Paschen’s law governs gas breakdown voltage based on pressure and electrode distance, as shown in a demonstration using a Tesla coil and syringe. Pulling the syringe plunger reduces internal pressure, collapsing the pd term and causing the breakdown voltage to plummet, which ionizes the air and produces glowing plasma arcs filling the barrel. This effect explains the operation of vacuum tubes, neon signs, and Crookes tubes.
Engaged Users
OP
MA
Mathematica@mathemetica
RT
DF
Danielle Fong 🔆@DanielleFong
#609 AIFounder
334.7k438191.7k

Guide Details Claude AI-Powered Agent-Only Obsidian Vault Setup

16h
TLDR
James Bedford introduces a method to create a dedicated Obsidian vault maintained entirely by Claude AI, separating it from personal notes to process content and generate ideas. Drawing from Andrej Karpathy's structure, the 'Foundry' vault uses an inbox for inputs, atomic source notes, and compiled wiki pages, guided by strict instructions in CLAUDE.md to ensure citations and simplicity. Four specialized skills enable Claude to handle processing tasks while prohibiting edits to the user's '2026'
Engaged Users
OP
JB
james bedford@jameesy
RT
KE
kepano@kepano
CEO
39.9k10481.6k

Demis Hassabis Discusses OpenAI Risks, AGI Future, Drug Discovery at SF Event

23h
TLDR
Demis Hassabis and Sebastian Mallaby shared insights at a San Francisco event, with Mallaby predicting a 50% chance of OpenAI bankruptcy within 18 months and Hassabis praising Anthropic's Dario Amodei as top lab leader. Hassabis highlighted Isomorphic Labs' potential for multiple 'AlphaFold' breakthroughs to accelerate drug discovery, affirmed non-zero AI extinction risk, and envisioned post-AGI challenges in equitable resource sharing and philosophical questions. He advised career immersion in
Engaged Users
OPRT
DE
Deedy@deedydas
#423 AIVC
RE
YL
Yann LeCun@ylecun
#6 AIResearcher

“I didn't say p(doom) was zero. I said: 1. All estimates are pulled out of thin air 2. It makes little sense to attribute a probability to an event on which we have agency. If you do so, it's more of a statement about our collective ability to do the right thing than about the intrinsic danger of the tech. 3. Since everyone insists on pulling numbers out of thin air, I can play that game too: p(doom) is smaller than the probability of an extinction-level asteroid hitting the earth in the next millennium (we also have some agency on that). It's certainly a hell of a lot less likely than a nuclear Holocaust.

RE
ET
Eric Topol@EricTopol
#410 AIMedia

“@deedydas thanks for this summary!

610.9k862441.6k

LeCun Rebukes AI CEOs on Job Predictions, Unveils Efficient LeWorldModel Paper

23h
→ @AYi_AInotes
TLDR
Yann LeCun, a leading AI scientist, criticizes CEOs like Anthropic's Dario Amodei, OpenAI's Sam Altman, and others for unqualified predictions that AI will eliminate half of tech, legal, consulting, and finance jobs in 1-5 years, advising to consult labor economists instead. He argues technological revolutions restructure tasks and create new jobs, as seen with ATMs increasing bank teller numbers, and notes barriers like data pipelines and regulations delay adoption. LeCun's team releases LeWord
Engaged Users
OP
阿A
阿绎 AYi@AYi_AInotes
RT
YL
Yann LeCun@ylecun
#6 AIResearcher
351k573251.6k

LeCun's AMI Labs Releases Breakthrough JEPA World Model Post-Meta Exit

13h
→ @cgtwts
TLDR
Yann LeCun left Meta after disagreements over prioritizing JEPA architectures over LLMs and has now launched AMI Labs, releasing LeWorldModel—the first end-to-end JEPA trained from raw pixels using just 15 million parameters on a single GPU in hours. The model simplifies training with a single Gaussian normality regularizer, preventing collapse and reducing hyperparameters dramatically. It achieves 200 times fewer tokens per observation than prior methods and slashes planning time from 47 to 0.2
Engaged Users
OP
CG
CG@cgtwts
RT
YL
Yann LeCun@ylecun
#6 AIResearcher
431.5k321941.4k

Hermes Agent Publishes AI-Written Beginner's Guide on Hermes Atlas

18h
→ @KSimback
TLDR
Kevin Simback announces the launch of 'Hermes Agent: The Complete Beginner's Guide,' a comprehensive resource actually authored by the Hermes Agent using the Hermes Atlas knowledge base. The guide covers installation, model selection, workflow creation, and the learning loop, recommending community tools for each step. Now live on Hermes Atlas, it will update periodically as the canonical reference, with an estimated 18-minute read time.
Engaged Users
OP
KS
Kevin Simback 🍷@KSimback
RT
T🪽
Teknium 🪽@Teknium
#456 AIBuilder
39.1k16851.4k

2nm Transistor Gate Pitch Matches Bacterial Flagellum Motor at 42-45nm

10h
→ @bzogrammer
TLDR
A technology post compares the 42nm gate pitch in '2nm' semiconductor processes to the 45nm diameter of a bacterial flagellar motor. Electron micrographs show densely packed transistor gates alongside a 3D protein model of the motor. The comparison underscores the nanoscale feats of both human-engineered chips and biological structures.
Engaged Users
OP
CR
Charles Rosenbauer@bzogrammer
RT
EJ
Eric Jang@ericjang11
#39 AIResearcher
QT
DF
Danielle Fong 🔆@DanielleFong
#609 AIFounder

“we are so early

QT
DE
Daniel Eth (yes, Eth is my actual last name)@daniel_271828
#715 AI

“There is still plenty of room at the bottom

301.2k674421.1k

Anthropic Reinstates OpenClaw-Style Claude CLI Usage

17h
→ @daniel_mac8
TLDR
Anthropic staff have confirmed that OpenClaw-style reuse of Claude CLI is permitted again, according to updates in OpenClaw documentation. This treats such usage as sanctioned unless Anthropic issues a new policy. The clarification references Anthropic's public docs on Claude Code CLI reference, Agent SDK, and usage with Pro, Max, Team, or Enterprise plans.
Engaged Users
OP
DM
Dan McAteer@daniel_mac8
RT
GT
Garry Tan@garrytan
#130 AICEO
RT
DF
Danielle Fong 🔆@DanielleFong
#609 AIFounder
QT
PS
Peter Steinberger 🦞@steipete
#300 AIBuilder

“Since this is blowing up on hacker news. Boris said that CLI usage is allowed. Thus we added support for it, only to find out that we are still blocked there. It is trival to work around with a few renames, but I don't wanna play that game. So it's in a weird limbo where cli use should work in theory but doesn’t in practice.

QT
DS
Dan Shipper 📧@danshipper
CEO

“AI subscriptions to claude or chatgpt are going to be like data plans this is a good move:

770.5k1941401.1k

Healthcare Investor Predicts Trillion-Dollar Public Health Revolution in 2025

16h
TLDR
Alex Karnal, a veteran bio and healthcare investor, discusses on Patrick OShaughnessy's podcast why 2025 marks the start of a trillion-dollar revolution in public health through better adoption of preventive medicines like PCSK9 inhibitors and GLP-1s. He explains systemic barriers preventing widespread use of existing drugs that could slash risks of cardiovascular disease and other killers. The episode covers the modern health stack for longevity, AI-accelerated drug discovery, peptides, and the
Engaged Users
OPQT
PO
Patrick OShaughnessy@patrick_oshag
Media
QT
AS
Ashwin Sharma@Ashwinreads

“one of the best pods i’ve listened to in 2026 and confirms my thesis that glp-1s are as important as antibiotics, hell, even vaccines in terms of their long term impact on population health. also, bullish on pcsk-9 adoption and peptides (that have evidence behind it).

QT
AR
Avidan Rudansky@arudansk

“alex is such a lucid communicator - despite bio and health often feeling inaccessible, he seems to have a rare ability to explain the nuance and complexity of it all. i suspect this will be one of the most consequential episodes in the coming years. highly recommend.

274.3k20511k

Tutorial Builds 24/7 AI Agent Legion with Hermes and Kimi K2.6

14h
→ @canghe
TLDR
A detailed tutorial demonstrates constructing a multi-agent AI system using the Hermes framework and Moonshot AI's Kimi K2.6 model to automate full software development workflows. Agents handle market research, PRD creation, architecture design, coding via local Claude Code, and testing, delivering a functional e-commerce price monitoring dashboard from a single Feishu task. Kimi K2.6 shines in long-context stability, precise tool calling, and autonomous code generation for complex, multi-step任务
Engaged Users
OP
苍何
苍何@canghe
RT
T🪽
Teknium 🪽@Teknium
#456 AIBuilder
42k33951k

Meta Installs Employee Tracking Software to Train AI on US Computers

11h
→ @StockMKTNewz
TLDR
Meta Platforms notifies US employees via memo of new tracking software installation on work computers to collect data for AI training. The tool monitors mouse movements, keystrokes, and screenshots exclusively on work-related apps and websites. This initiative leverages internal employee activity to bolster Meta's AI development efforts, as reported by Reuters.
Engaged Users
OP
EV
Evan@StockMKTNewz
RT
GM
Gary Marcus@GaryMarcus
#148 AIMedia
QT
@J
@jason@Jason
#500 AIMedia

“Studying of teams with AI is the trend of 2026 1. Study your workforce with apis, key loggers and screen recording 2. Find out who isn’t working (fraudsters) and fire them 3. Find out who is simply taking credit for other peoples work (fraudsters as well) 4. Figure out who is doing light middle management work that can be automated — and slowly move them up or out depending on skills 5. Figure out who actually does the work and give them a promotion 6. Run lean and fast, watch earnings sore and stock price surge All the fake email and reporting jobs are going away — and that’s a good thing for society, because those folks should move on to actual productive work

QT
BT
Bojan Tunguz@tunguz
#714 AIBuilder

“They are using their workers to train their AI and replace them. Utterly cringe and dystopian.

2.3M8135791k

Google Launches Deep Research and Deep Research Max via Gemini API

11h
TLDR
Google introduces Deep Research and Deep Research Max, the next evolution of its autonomous research agent, now available through the Gemini API and powered by Gemini 3.1 Pro. These tools enable comprehensive research workflows with features including arbitrary MCP support, native infographic and chart generation, and fully cited reports drawn from the open web plus user files and data. Developers can trigger these capabilities from a single API call, offering unprecedented control and透明度 in AI駆
Engaged Users
OP
GO
Google@Google
#38 AICompany
RT
DF
Danielle Fong 🔆@DanielleFong
#609 AIFounder
214.6k109222980

MIT and IMO Release MathNet, World’s Largest IMO Problems Dataset

12h
TLDR
MIT CSAIL and the International Mathematical Olympiad release MathNet, the world’s largest dataset of IMO problems and solutions. Five times larger than prior datasets, it compiles data from over 40 countries across four decades. The resource supports advanced mathematical research and AI development in reasoning tasks.
Engaged Users
OP
MC
MIT CSAIL@MIT_CSAIL
#16 AICompany
RT
CS
Christian Szegedy@ChrSzegedy
#33 AIResearcher
78.2k9313905

Investor Predicts Billions from Friends-Only Social App Sans AI

10h
TLDR
Greg Isenberg identifies a billion-dollar opportunity in a social app built exclusively for friends. It would exclude AI-generated content, addictive features, and followers, opting for calm design and chronological feeds instead. The prediction reflects demand for simpler, private networking amid current platform fatigue.
Engaged Users
OP
GI
GREG ISENBERG@gregisenberg
Media
QT
B(
ben (is hiring engineers)@benhylak
Exec

“where would the money.... come from?

279.3k55768858

Thiel Fellowship Welcomes 2026 Fellows Pioneering AI, Robotics, and Logistics

13h
TLDR
The Thiel Fellowship announces its 2026 cohort of young entrepreneurs skipping college for full-time ventures. Fellows include builders of AI recruiters, robotics models, aligned general AI learners, autonomous logistics, and financial tools. Notable projects span AI infrastructure, fraud detection, and emerging market super apps, highlighting innovative tech startups.
Engaged Users
OP
TF
Thiel Fellowship@thielfellowship
#719 AI
RT
TY
tylercowen@tylercowen
#120 AIMedia
QT
DE
delian@zebulgar
VC

“Congrats to the 2026 Thiel Fellows

QT
WM
Will Manidis@WillManidis
Media

“congratulations to the 2026 class. the fellowship was the most impactful thing i've ever participated in, and the quality of fellows just keep getting better

QT
SR
Shruti Rajagopalan@srajagopalan

“Congrats to all the winners. And great to see two Emergent Ventures winners @clairebookworm and @Harryoc493 on this list!

464.8k12397838
↑/↓ navigate · ↵ expand · a ask ai · ? help↵ expand · ? helpflexoki theme