Ex-Anthropic, OpenAI, DeepMind and xAI researchers just raised $1B for Mirendil, to automate AI research itself.
Its base model took 3.5 months and under $10M in compute.
The next race isn't building models. It's building what builds them.
#AI
https://t.co/EdTefZiJNH
Jensen Huang, 33 yıl sonra Nvidia CEO'luğundan istifa etti. Geçici CEO: CFO Colette Kress. Sebep açıklanmadı.
Sözün bitti yer 😵💫
#YapayZeka#Nvidia@nvidia
https://t.co/ernEhuNO1X
Anthropic tested 4 multi-agent setups: one agent per user did 10-45 pts worse than one shared agent.
Not dumb agents. Late ones, missing each other's messages.
More agents ≠ better outcomes. Coordination is the product.
#AI#AIAgents@AnthropicAI
https://t.co/zq3yDvabeU
@GoogleDeepMind's agents formally verified Signal's core crypto code in just 4 weeks, writing millions of lines of Lean proofs. Humans reviewed ~1% of it.
The real shift: #agents aren't just writing code anymore, they're proving it's correct.
#AI
https://t.co/N5SKsaRIXF
I just had the best Turkish tea and simit with volleyball legend @edaerdem14. Thanks for the wonderful conversation and warm welcome to beautiful Ankara!
Alibaba'nın yeni testinde en iyi kodlama ajanı bile gerçek kullanıcı taleplerinin sadece %26,4'ünü eksiksiz çözebildi. Eski benchmark'larda skorlar %99'un üzerinde.
Ajan seçerken benchmark'a değil, kendi iş akışına bak.
#YapayZeka
https://t.co/PDvjY9m6V4
Satya Nadella pictures @Microsoft with 200K employees and "20 million agents all in a loop."
The hard part isn't building 20M agents. It's deciding which can act alone and which must hand off to a human.
Scale without that line is scaled risk.
#AIAgents
https://t.co/yuaXoVGOHv
AI ajanları kullanıcı adına iş yaptıkça asıl soru şu: hangi ajan kimin adına hareket edebilir ve nasıl durdurulur? @awscloud ve @okta'nın cevabı: her ajana ayrı kimlik, her araca ayrı izin, her karara kayıt. 🔐
#YapayZeka#AIAgents
https://t.co/7WdVdv8qls
Greenlight'ın AI ajanları ayda ~4.000 güvenlik vakasının %90'ından fazlasını insansız çözüyor.
Asıl başarı tasarımda: emin olmayan ajan vakayı üst katmana, riskliyi insana devrediyor.
İyi ajan her şeyi yapan değil, devretme anını bilen...
#YapayZeka
https://t.co/Qa87tgFQJY
@awscloud just brought @nvidia MIG to SageMaker Inference: one GPU split into up to 7 isolated slices, each running its own model.
If GPU cost is the real blocker for on-prem AI, this opens a serious door. 🧠
#AI#GPU
https://t.co/uat7uTLOQN
You only need 4 questions to design an AI agent:
When? → Trigger
What? → Agent
Who? → Skill
How? → Context
The Amazon Quick team's framework makes it obvious what's missing in your next automation. 💡
#AI#AgenticAI
https://t.co/OwswiCOJhH
I imagine an ecosystem made up of people who are inspired not by others’ failures, but by others’ happiness and success. Naive? Maybe.Sora was shut down. And some people celebrated. When I saw it, I paused. It felt strange. OpenAI made a decision. They redirected resources to coding and agent systems. It’s a rational business move. Sora isn’t gone, it’s just been put aside for now. Most likely, it will come back in another form, in another place.
But some people treated this like a victory.
“Hahaha, see? It didn’t work.”
“I told you so.”
“AI needs to know its limits.”
And so on…And this is exactly where I got stuck. Not because I’m a fan of Sora, but because of what this says sociologically. Because this isn’t about Sora. I’ve been seeing this pattern everywhere for the past few years. Something fails, something stumbles, and a group of people immediately feels relief. As if others’ failure or unhappiness has become a need.When did we become like this?
If anyone knows the starting point or the reason, please tell me :) There are people who feel like they’re moving forward when someone else stumbles. But this kind of satisfaction doesn’t take anyone anywhere. Someone who celebrates Sora’s shutdown gains nothing about the future of video AI. They just feel a momentary relief. Relief is one thing, progress is another.And those who can’t see the difference will keep celebrating every failure.
But when they look around, they’ll realize they’re still standing in the exact same place.
We’re saying goodbye to the Sora app. To everyone who created with Sora, shared it, and built community around it: thank you. What you made with Sora mattered, and we know this news is disappointing.
We’ll share more soon, including timelines for the app and API and details on preserving your work. – The Sora Team
@soraofficialapp I imagine an ecosystem made up of people who are inspired not by others’ failures, but by others’ happiness and success. Naive? Maybe.
Sora was shut down. And some people celebrated. When I saw it,it felt strange. Why?
Honestly, #Intel#ArcProB70 feels like more than a benchmark win.
At $949 with 32GB VRAM, it makes running #LLM models like #Llama3 and #Qwen locally far more realistic—shifting toward #LocalAI with better privacy, control, and lower-cost #GenAI.
@pmitu The real break won’t be about technology. It will be about society. The world will split into two: macro societies that produce technology, and micro societies that are forced to use it. That’s where the real challenge will begin. 🤷🏻♀️
Office life before the invention of AutoCAD and other drafting software.
Prior to the release of AutoCAD in 1982, engineering drawings were all done by hand using different grade pencils, erasers, T-squares and set squares.
Nobel Prize–winning economist @JosephEStiglitz warns that #AI-generated content could degrade the global information ecosystem.
But As #GenerativeAI floods the internet with synthetic text, human-created knowledge may become the most valuable data online.
AI is moving beyond chat — we’re entering the era of #AIAgents.
New developments from companies like #Microsoft and #Datadog show that #LLM-powered agents can now access real-time data, execute workflows, and operate across tools.
The shift:
#AIAssistants → #AutonomousAgents