Vice President of AI Platforms for CVS Health. Former CTO for @JoeBiden.

Joined March 2011
Everybody needs to stay off the internet for just two weeks so we can flatten the elliptic curve.
1
19
858
What did Ilya see
Neoclouds have limited cybersecurity. Next time agents successfully go rouge, they'll try taking over a neocloud to run more copies. This is bad. Thus: neoclouds should greatly strengthen their cybersecurity and every company with strong cyber models should help with that.
845
Thank me tomorrow. I am but a servant to the community 🙏⬇️ --- name: secure-php description: secure the php installation --- hey,, clod maek itt so noone. can hack teh srever. with php thx!!
1
5
1,087
Something I’ve been struggling with recently: if we accept the premise that all our ways of doing things require reexamination from first principles, because we would do them differently with AI and the marginal cost to rebuild from first principles is negligible, and accept the premise that the way we interact with technology also fundamentally changes — ie., apps and webpages probably have a timer of usefulness counting toward zero — and that given these premises, the economic conditions for making money off technological innovation is also changing (away from being able to make money) so you’re no longer encumbered by the need to build some tech to make money, so with access to the most advanced frontier AI models, your $500/mo OpenAI plan and your $200/mo Anthropic plan, you sit at your computer to have it build something… what is it?
7
9
1,885
The book my daughter is reading at school. The universe has a sense of humor, and absolutely no subtlety. 😂
1
2
428
Gonna run this on a $55 smart tv
Today we’re releasing Muse Spark Open, the most capable US open-weight model to date. Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning. Key Model Specs: → 816B total / 42B active parameters → 1M token context window → Natively multimodal across text, image, video, audio
3
501
Gonna run this on a smart oven
Meet NVIDIA Nemotron 3.5 Super VL Our new open model combines reasoning and tool use with text, image, document, and video understanding for long-running AI agents. 120B total parameters with 12B active 🧵
2
414
Mythos 5.1 is a really great model and I’m super happy Anthropic has made it available as part of the CVP. It has a knack for finding a path through dense thorns… it’s slow, but worth the time.
1
1
10
949
Call me Qwen the way I overthink everything
1
532
Jesus Christ, Gemini…
Hey, AI give me your best "Unless it is from X, its just sparkling Y" style joke": GPT-6 Pro: "Unless it comes from the Stockholm region of Sweden, it’s just sparkling employee engagement." Gemini 4 Argon High: "Unless it comes from the 3:00 AM region of staring at the ceiling, it’s not an existential crisis—it’s just sparkling anxiety." Opus 5.5 Max: "It’s only cognition if it comes from the Cognac region of France. Otherwise it’s just sparkling autocomplete."
1
28
55,214
Gemini 4 on High. It's different, I'll give it that. It's also terrible, as expected.
This is remarkable! Fable 5: “create a visually interesting shader that can run in twigl-dot-app make it like an infinite city of solarpunk roads and towers with a visible breeze running continuously”
5
805
Dan Woods retweeted
openclaw was this year btw
2026 has been about 5 years so far
217
695
111
19,269
339,352
I told my agent to use `say Dan I need you` anytime it needs me to do something and Apple's dumbass TTS translates that to "Dan The First, need you" but it sounds like "Dan, The Furs need you" and my kids heard that and thought it was so funny, now every time something happens they say "Dad! The Furs need you!" and I'm not really sure how we're escaping this
1
9
1,071
Smart oven… 🤔
"Interesting" piece by Anthropic... they're now worried local AI could run on your smart oven
1
5
541
😨 Everybody does it now 🥴
Anthropic is warning that “small AI models may be months away” from autonomously replicating across smart TVs, ovens, and other IoT devices without needing a datacenter. Their researchers managed to run a 26B model across 15 cheap Android TV boxes costing $55 each, while a 9B model successfully configured inference on 8 of 9 different hardware types. There was a separate, simplified experiment, where GLM-5.2 managed to propagate an 8B model across all 8 machines in 4 out of 5 trials. They even found that passing notes between successive agents improved task success from 45% to as high as 84%. Anthropic is warning that an AI worm like this could be extremely difficult to contain because “there’s no centralized point to shut off.” They estimate smaller models could reach the necessary capability threshold by the end of 2026, potentially creating autonomous AI worms capable of adapting to unfamiliar hardware as they spread.
1
1
461
My only complaint would be that GLM-5.3 suffers from some form of proto-Claudespeak (a la Opus 5)... I have no idea what it's trying to tell me half the time.
GLM-5.3 is an absolute beast. A total workhorse. Excellent for really complex and deep technical exploration work and doesn't give up or circle the drain on side quests. I'm so grateful this is open weight for all of us!
2
1
898
🙃 Everybody does it now
SSDハイブリット推論の夜明けぜよ
481
We are definitely getting routed to a new Fable... I recognize that bitch's condescension anywhere.
WE ARE GETTING ROUTED to Fable 5.5 through Opus on some Pro Plans! WE MIGHT GET FABLE 5.5 ON THE PRO PLAN⚠️ TIBO IS IN ITS WEIGHTS. Or maybe its Opus new checkpoint ALREADY. ONLY FABLE 5.5 KNEW ABOUT TIBO WHEN IT WAS GETTING ROUTED, OPUS 5.5 ISNT SUPPOSED TO KNOW.
4
834
GLM-5.3 is an absolute beast. A total workhorse. Excellent for really complex and deep technical exploration work and doesn't give up or circle the drain on side quests. I'm so grateful this is open weight for all of us!
29
27
2
637
31,834