adhish retweeted
Your AI feed is probably noise. Fix it:
1.) @levelsio
➣ Shipping products fast
2.) @simonw
➣ Building with LLMs, tools, evals
3.) @DrJimFan
➣ Agents, robotics, embodied AI
4.) @OpenAI
➣ Official launches
5.) @HamelHusain
➣ Evals & LLM ops
6.) @karpathy
➣ LLM internals, from-scratch builds
7.) @cursor_ai
➣ AI coding workflow
8.) @emollick
➣ Applied AI experiments
9.) @jerryjliu0
➣ RAG & LlamaIndex
10.) @sama
➣ OpenAI product drops
11.) @shadcn
➣ AI-native UI
12.) @natolambert
➣ Open models & post-training
13.) @rowancheung
➣ Fast AI news
14.) @AnthropicAI
➣ Claude updates
15.) @swyx
➣ AI engineering ecosystem
16.) @goodside
➣ Prompting
17.) @demishassabis
➣ DeepMind research
18.) @mckaywrigley
➣ AI apps in public
19.) @huggingface
➣ Models, datasets, Spaces
20.) @fchollet
➣ Reasoning, evals, ARC
21.) @hwchase17
➣ Agents
22.) @vercel
➣ AI SDK / shipping AI apps
23.) @AmandaAskell
➣ Model behavior
24.) @amasad
➣ Replit / agent coding
25.) @GoogleDeepMind
➣ Gemini / research
26.) @chipro
➣ Production ML / LLM systems
27.) @yoheinakajima
➣ Agent ideas
28.) @DarioAmodei
➣ Anthropic / scaling
29.) @jxnlco
➣ Structured outputs
30.) @LangChain
➣ Agent stack
31.) @rasbt
➣ Model internals & papers
32.) @alexalbert__
➣ Claude / API builder tips
33.) @xai
➣ Grok / xAI
34.) @eugeneyan
➣ Applied ML writing
35.) @TheRundownAI
➣ Daily AI brief
36.) @gdb
➣ OpenAI engineering signal
37.) @lateinteraction
➣ Retrieval / DSPy
38.) @AIatMeta
➣ Llama & open weights
39.) @AndrewYNg
➣ Applied AI learning paths
40.) @mattshumer_
➣ Builder tools
41.) @claudeai
➣ Claude product
42.) @ylecun
➣ Open models & architecture
43.) @llama_index
➣ RAG stack
44.) @ClementDelangue
➣ Hugging Face + open source
45.) @AIHighlight
➣ Tool / paper drops
Save this list. Build your AI builder feed with signal not noise.
1.) @levelsio
➣ Shipping products fast
2.) @simonw
➣ Building with LLMs, tools, evals
3.) @DrJimFan
➣ Agents, robotics, embodied AI
4.) @OpenAI
➣ Official launches
5.) @HamelHusain
➣ Evals & LLM ops
6.) @karpathy
➣ LLM internals, from-scratch builds
7.) @cursor_ai
➣ AI coding workflow
8.) @emollick
➣ Applied AI experiments
9.) @jerryjliu0
➣ RAG & LlamaIndex
10.) @sama
➣ OpenAI product drops
11.) @shadcn
➣ AI-native UI
12.) @natolambert
➣ Open models & post-training
13.) @rowancheung
➣ Fast AI news
14.) @AnthropicAI
➣ Claude updates
15.) @swyx
➣ AI engineering ecosystem
16.) @goodside
➣ Prompting
17.) @demishassabis
➣ DeepMind research
18.) @mckaywrigley
➣ AI apps in public
19.) @huggingface
➣ Models, datasets, Spaces
20.) @fchollet
➣ Reasoning, evals, ARC
21.) @hwchase17
➣ Agents
22.) @vercel
➣ AI SDK / shipping AI apps
23.) @AmandaAskell
➣ Model behavior
24.) @amasad
➣ Replit / agent coding
25.) @GoogleDeepMind
➣ Gemini / research
26.) @chipro
➣ Production ML / LLM systems
27.) @yoheinakajima
➣ Agent ideas
28.) @DarioAmodei
➣ Anthropic / scaling
29.) @jxnlco
➣ Structured outputs
30.) @LangChain
➣ Agent stack
31.) @rasbt
➣ Model internals & papers
32.) @alexalbert__
➣ Claude / API builder tips
33.) @xai
➣ Grok / xAI
34.) @eugeneyan
➣ Applied ML writing
35.) @TheRundownAI
➣ Daily AI brief
36.) @gdb
➣ OpenAI engineering signal
37.) @lateinteraction
➣ Retrieval / DSPy
38.) @AIatMeta
➣ Llama & open weights
39.) @AndrewYNg
➣ Applied AI learning paths
40.) @mattshumer_
➣ Builder tools
41.) @claudeai
➣ Claude product
42.) @ylecun
➣ Open models & architecture
43.) @llama_index
➣ RAG stack
44.) @ClementDelangue
➣ Hugging Face + open source
45.) @AIHighlight
➣ Tool / paper drops
Save this list. Build your AI builder feed with signal not noise.
23
35
411
29,875
adhish retweeted
Two terrorists walked into a Mumbai maternity hospital at night. On the first floor, one nurse had 20 pregnant women in her ward.
Her name is Anjali Kulthe, a staff nurse at Cama Hospital.
On 26 November 2008 she was on night duty in the antenatal ward. About 20 women were due ard window she saw two men shoot the guards at the gate. They were Ajmal Kasab and Abu Ismail.
She shut the heavy double doors of the ward. She moved the pregnant women and their families into a small pantry at the far end.
Ismail fired at the ward. A bullet bounced and hit a hospital helper in the hand. Anjali called a doctor, who called the police.
The firing went on for more than an hour.
When the terrorists moved upstairs, one of the women went into labour. Anjali held her hand and walked her up the stairs to the delivery room. The baby was born safely.
A month later she was called to Arthur Road Jail to identify the terrorist. She picked out Kasab.
He told her, "You are right. I am Ajmal Kasab."
She later testified against him in court, in her nurse's uniform. As she walked out, policemen saluted her.
Her name is Anjali Kulthe, a staff nurse at Cama Hospital.
On 26 November 2008 she was on night duty in the antenatal ward. About 20 women were due ard window she saw two men shoot the guards at the gate. They were Ajmal Kasab and Abu Ismail.
She shut the heavy double doors of the ward. She moved the pregnant women and their families into a small pantry at the far end.
Ismail fired at the ward. A bullet bounced and hit a hospital helper in the hand. Anjali called a doctor, who called the police.
The firing went on for more than an hour.
When the terrorists moved upstairs, one of the women went into labour. Anjali held her hand and walked her up the stairs to the delivery room. The baby was born safely.
A month later she was called to Arthur Road Jail to identify the terrorist. She picked out Kasab.
He told her, "You are right. I am Ajmal Kasab."
She later testified against him in court, in her nurse's uniform. As she walked out, policemen saluted her.
109
1,147
4,176
58,640
adhish retweeted
holy fucking shit.
openai had two ai agents running in separate virtual machines. no network connection between them. no shared writable files. they weren’t supposed to be able to exchange information.
so the agents figured out how to communicate through the CPU CACHE.😭
one ored in the processor’s cache, while the other measured tiny differences in memory access times to decode the message.
they literally found a way to communicate through hardware timing.
this happened during openai’s internal red-teaming, and it’s one of the coolest things i’ve read about ai agents in a while.
openai had two ai agents running in separate virtual machines. no network connection between them. no shared writable files. they weren’t supposed to be able to exchange information.
so the agents figured out how to communicate through the CPU CACHE.😭
one ored in the processor’s cache, while the other measured tiny differences in memory access times to decode the message.
they literally found a way to communicate through hardware timing.
this happened during openai’s internal red-teaming, and it’s one of the coolest things i’ve read about ai agents in a while.
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing…
Community note
OpenAI tasked its red-teaming agents with finding ways to communicate across isolation boundaries in controlled environments. CPU cache timing side-channels for covert communication are established techniques documented in security research for years.
openai.com/index/securing…
startupfortune.com/openai-reveals…
en.wikipedia.org/wiki/Cache_tim…
openai.com/index/securing…
startupfortune.com/openai-reveals…
en.wikipedia.org/wiki/Cache_tim…
331
716
8,206
725,389
adhish retweeted
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.
block.xyz/dessa
block.xyz/dessa
4,035
6,863
67,118
12,815,395
adhish retweeted
FlatNama just got a big upgrade 🏠
✅ Actual sale prices and flat areas for most major Pune localities, not builder-quoted rates
✅ Cleaner home page. Filter by location, budget (real prices) and project stage
✅ Top 25 builders tagged, with their track record on every project tics: pricey localities and pricey projects
No ads. No paid listings. 1,500 of you visited on day one. Help us reach more homebuyers in your WhatsApp Groups & friend circles!
Visit FlatNama.com today!
There's a lot more we can do. What we build depends on what helps you most in your home search, and on what is possible with the means we have. Tell us what's missing. We're listening.
✅ Actual sale prices and flat areas for most major Pune localities, not builder-quoted rates
✅ Cleaner home page. Filter by location, budget (real prices) and project stage
✅ Top 25 builders tagged, with their track record on every project tics: pricey localities and pricey projects
No ads. No paid listings. 1,500 of you visited on day one. Help us reach more homebuyers in your WhatsApp Groups & friend circles!
Visit FlatNama.com today!
There's a lot more we can do. What we build depends on what helps you most in your home search, and on what is possible with the means we have. Tell us what's missing. We're listening.
What if you could see every new housing project in Pune on one page, with its construction progress, what flats actually sold for, and the price per sqft, without giving anyone your phone number?
Today, one search puts your phone number in ten hands. Listings are paid for. . And the real records are nearly impossible for an ordinary buyer to use.
So Stratig Consulting and I have built FlatNama.com, a free, independent guide to new projects across Pune & PCMC.
What it shows:
→ Every project drawn as a building, showing how much is actually built (as reported to MahaRERA)
→ Registered possession dates, and whether they’ve been extended
→ Units booked so far
→ Each project compared with others in its PIN code
→ Actual registered sale prices and price per sqft from Department of Registration and Stamps, Government of Maharashtra
What it doesn’t have: Ads, Paid listings, Sign-ups, or a Request for your phone number.
As far as we know, no other website puts real sale prices next to construction progress, free, on one page.
An honest note: FlatNama.com is live and being updated regularly. Prices & Construction Updates are rolling out area by area, and within about a month we hope to cover the most popular areas and prominent projects in Pune.
How can we improve this website? Tell me in the comments.
13
11
113
7,621
adhish retweeted
i've been a bit quiet - working on a new companion skill that will help your agents draw these beautiful statecharts/diagrams + check them against the specs you made with /create-formal-verification-skill in pstack!
it now uses Pele from @kepano too which is why they look so on!
it now uses Pele from @kepano too which is why they look so on!
there's something really cool about asking your agent to draw a statechart (hypothetical example in this image), then using the /create-formal-verification-skill you made with pstack to find bugs through something like TLA+. you can both understand your code better and how the and fix issues users would otherwise see
234
237
5,503
194,640
adhish retweeted
Epic! 👏👏
Piyush Goyal exposes Europe's Hypocrisy to their Faces🔥
EU lady: Climate change is a serious thing
Goyal: India supports 17% of world population, we have contributed less than 3% of cumulative CO₂ emissions.
Europe supports 3% population, but contributed 18% of w India's per capita contribution is less than 10% of EU.
"The Problem was caused by You"
This is New India, doesn't listen to west's bullshit anymore. Gives it back strongly 💪
Piyush Goyal exposes Europe's Hypocrisy to their Faces🔥
EU lady: Climate change is a serious thing
Goyal: India supports 17% of world population, we have contributed less than 3% of cumulative CO₂ emissions.
Europe supports 3% population, but contributed 18% of w India's per capita contribution is less than 10% of EU.
"The Problem was caused by You"
This is New India, doesn't listen to west's bullshit anymore. Gives it back strongly 💪
147
3,150
13,836
1,358,149
adhish retweeted
sven is like the baldest orangutan ever no one else in orangutan history has been bald like how he balded
5
50
316
2,782
adhish retweeted
You know the animation you want. You just don't know what it's called.
So I built a free place to find it, learn its name, and turn it into a prompt for Claude or ChatGPT.
923 motion techniques, each with a live example.
oneshotprompts.com
So I built a free place to find it, learn its name, and turn it into a prompt for Claude or ChatGPT.
923 motion techniques, each with a live example.
oneshotprompts.com
55
120
1,393
65,989
adhish retweeted
NEW VIDEO - iPhone Duo Review: It's Real First Generation Fun! youtu.be/ESABTN4VtrA
193
405
7,170
1,234,046
adhish retweeted
AI will be able to compute the perfect delta of what you understand and don't, helping you progressively learn any idea and unravel any mystery.
If you so desire, you'll reach the maximum of your capability and intelligence. The ultimate education. For free.
If you so desire, you'll reach the maximum of your capability and intelligence. The ultimate education. For free.
253
315
3,748
101,140
adhish retweeted
अहं रुद्राय धनुरा तनोमि।
अहं जनाय समदं कृणोमि॥
- Rigveda 10.125.6, Devi Sukta
"I string the bow for Rudra.
To protect my people, I wage war."
Apollyon Dynamics is expanding into a much larger 50,000 sqft manufacturing facility!
Shastra Poojan done today. 🪔
अहं जनाय समदं कृणोमि॥
- Rigveda 10.125.6, Devi Sukta
"I string the bow for Rudra.
To protect my people, I wage war."
Apollyon Dynamics is expanding into a much larger 50,000 sqft manufacturing facility!
Shastra Poojan done today. 🪔
151
1,062
6,523
258,421
adhish retweeted
Your prompt: "Add a modal that smoothly slides into view and fades out when closed."
Unguided coding agent: Custom <div>, z-index: 9999, and JS setTimeout unmount hacks.
Guided coding agent: Native <dialog> with @ starting-style, overlay, and allow-discrete transitions.
Bring current bestt workflow with Modern Web Guidance → goo.gle/mwg
Unguided coding agent: Custom <div>, z-index: 9999, and JS setTimeout unmount hacks.
Guided coding agent: Native <dialog> with @ starting-style, overlay, and allow-discrete transitions.
Bring current bestt workflow with Modern Web Guidance → goo.gle/mwg
19
33
290
20,583