AI-powered builder, fitness nerd, travel enthusiast, and madridista

Austin, TX
Joined March 2020
Interesting that claude now has ultracode as separate and independent from the effort setting. This makes a lot more sense imo
1
34
Rishi Venkat retweeted
Instinct on the Action Button You can now make Instinct your iPhone’s Action Button. Press and hold, say what you need, and Instinct will take it from there, without you needing to leave the app you’re on. To set it up: 1. Go to app.instinct.com/voice-short… and tap Add Shortcut 2. Go to Settings → Action Button → Shortcut → Instinct. This is available to all Instinct users on an iPhone 15 Pro and newer.
118
42
40
1,108
192,658
Just imagine 2016/17 Madrid against this Barcelona... Ronaldo would have had 40 more el clasico goals
36
Rishi Venkat retweeted
Gemini 4 Argon is our next era of frontier intelligence. It shows significant improvements across benchmarks, setting a new state of the art for real-world long-horizon software engineering tasks.
206
541
675
6,677
2,071,898
Rishi Venkat retweeted
Three thoughts after playing with Jev: - this is way more than a simplistic classifier: you can pack a lot of complex state and do more than you’d think - it’ll take a minute to think about your problems in a non-LLM shape - not quite beating frontier LLMs in correctness most of the time so can’t quite switch yet but it’ll clearly get there - this is definitely a new model type we will all use by next year; I wouldn’t ignore it
40
24
6
542
52,476
Rishi Venkat retweeted
Third iteration with Opus 5.5 medium. I am in awe. It turned blender, image-gen, and three.js into this beauty, which runs on your browser. An understatement to say that Anthropic cooked. You are looking at the future of game dev.
480
513
236
8,957
1,842,547
If you’re like me, you have 1,000+ threads across Codex and Claude Code, and your agent takes forever to find the one it needs. I made Jev Thread Search (link below) to search your local Codex conversations by describing what you remember. It's blazing fast and < $.01/search.
1
2
106
Introducing ContentBlocker, a Jev-powered browser extension that temporarily blocks content. You could say “hide spoilers for the show I’m watching" or “I only want to see content about the GPT 6 and Opus 5.5 releases” and ContentBlocker will filter the content in real time.
1
63
Rishi Venkat retweeted
BRO. Jev just opened to everyone and I've been testing it ALL day - it is actually nuts. This feels exactly like GPT-3.5 launch week. Everyone's building, nobody fully understands it yet, and honestly it's crazy how fast this is moving. If you're not experimenting with this right now you're missing the window. People are already running browser agents for a tenth of a cent per task, sorting thousands of emails, making trading calls in milliseconds, scoring Meta ads to find winners before burning budget... The model cannot write a single word. It only decides. That's the entire trick. I'm compiling everything I learned into an article right now, dropping it SOON.
Jev is now available to everyone. No waitlist. Start using it here: console.typesafe.ai
10
13
1
285
71,245
Rishi Venkat retweeted
Our job is clear now: build System 1.5 Connect System 1 (Jev) to System 2 (frontier reasoning models) using really good software architecture and deterministic connective tissue.
67
36
17
634
78,306
Diffusion LLMs and speculative decoding promise much faster agents. Yet agents need long contexts, and training on them is painfully slow. Introducing Context-Sharded Block Parallelism (CSBP), a new distributed parallelism strategy unlocking significant training efficiency for diffusion LLMs, with speedup gains growing with context length 🚀 ⚡ 7.59× faster DFlash2 speculative decoding drafter training ⚡ 1.61× faster block diffusion fine-tuning ⚡ 1.33× faster autoregressive → block diffusion adaptation With the same GPU hours, models trained with CSBP score higher on SWE-bench Verified and Terminal-Bench Lite 👑 Open-sourced in Turbo-dLLM, our new optimized distributed training library. Advised by @Azaliamirh and with an amazing team: @PranshuChatur11 @hangoo_kang @pshroff_ @ishanskhare @KumbongHermann
10
35
9
150
83,110
Rishi Venkat retweeted
How I feel every time I copy something from my iPhone and paste it on my Macbook
55
805
95
13,974
460,626
I've been using Instinct a ton more after adding a shortcut to talk to it from the Action Button. Here's the one I built: tinyurl.com/5e9jdbjm @noahrshinn this should come as an onboarding recommendation for Instinct - the voice interaction model is a lot more convenient.
1
1,174
Yesterday I spent the day at the Imagination in Action AI Summit at Google's Bay View campus. Between the stages and the demo floor, three ideas stuck with me 🧵:
3
49
🧵(3/3) We have no idea how the brain mechanistically works - studies have only produced correlations. AI is in no way replicative of actual mammalian intelligence, even if it may have the same outcomes.
7
(🧵 2/3) From Chirag Mahapatra (Mercor): Data is the bottleneck, and physical data is the next frontier. Real world environments and workflows are tough to source, but a good approximation is critical to create high quality data to push the frontier.
15
🧵(1/3) Application-layer companies will thrive ... IF they adapt. People have been concerned about Anthropic and OpenAI eating up all the opportunity, but there will always be room for builders who focus on a specific problem and solve it better than anyone else.
32
Rishi Venkat retweeted
Much has been said by racists, politicians, and racist politicians (like Bo French) about Indians and UT Austin admissions. I thought I should shed some light so their racism could be slightly more intelligent. I'm a brown kid. I grew up in Frisco, went to a massive Texas high school, and went to UT. I would wager that the kids in this picture and I have practically the same background. Like me, they probably went to a massive, pressure-cooker suburban Texas high school, and survived the gauntlet to earn their admissions. These aren’t foreigners. State law requires 90% of UT's freshman class to be Texas residents. 75% of those in-state seats go to automatic admission on class rank. These are Texans. Texas, for better or worse, is one of the few meritocratic college admissions schemes in the country for a school of its caliber. The rules are simple. Today, the top 5% of a public school's graduating class is automatically accepted. There is nothing to game. You can’t fake being a competitive rower or have some consultant write your essays for you. There isn’t anything to steal; it’s all there on the table to compete for. And the competition starts the moment the bell rings for your first period of freshman year. You are ranked in your class every single year, for us, #1 to #450. You knew exactly where you were, exactly how hard you had to compete. My year was even less competitive than it is today. The top 6% made the cut, so that’s an all-out war for ~27 kids to get auto-accepted into UT. So we worked our butts off. Here's what my life was like, starting in kindergarten. Every summer, before the school year started, I had to finish the upcoming year's math textbook. I sweated over getting As, even in elementary school. One funny memory is walking to my parents' room, crying, to report my first-ever 89 on a report card. I did Kumon, my least favorite thing in the world. I still hate that weird half-smile that is their logo. I went to math and science competitions, where I always lost, and came home crying again. (Quite the crybaby, wasn't I!) And everyone else around me worked just as hard. Actually, a lot of the kids worked much harder. I know people who took extra classes (zero period) just to boost their GPA. Who routinely slept 5 hours a night to nail every assignment they had after band practice. Who were distraught when they got a 98 instead of a 100. People who stopped doing things they loved, like sports or arts, because those classes dragged down their GPA. People put their entire life force into getting into this school. I wasn't even in the top 10% of my class. The competition was too good. I wasn't disciplined enough to hang with them. I got in as one of the 22.5% of admits who are at-large Texans. These are American kids who earned their way into this school in the most American way possible. Competition. They were told the rules, they competed, and they got the number. And on Saturday they were in that stadium in burnt orange, screaming their heads off, like every other Texan at DKR or around the country. I can’t feel anything but pride for them. That they get to live the dream that so many of us dreamt growing up! Kids like this have helped improve UT’s standing on the national stage. They made it out of the pressure cooker and placed well in Finance, Big Tech, and more recently, startups. They meaningfully contribute to the economies of our state and country. You should be proud of them too. Overall, this seems like a meritocratic (albeit flawed) way for Texans to be admitted to our state’s flagship school. It’s racially blind. But politicians like Bo French are asking for something different. They want the student body to rebalance racially. Surely they aren’t advocating for affirmative action, are they?
I heard UT graduation this year looked like this. I didn’t believe it. The problem is now obviously far worse than anyone imagined.
1,903
1,273
355
9,858
1,656,003