Founder of SafeAI, autonomous heavy industry. Excited about AI and the physical world

Sunnyvale, CA
A startup has 99 ways to die. If you are going to build one, don’t play for the small thing? As you get more experience, maybe you bring that 99 down to 80 or 70. But the risk never goes away. If you are in Silicon Valley, or the US for that matter, you are lucky to be here. Take a bigger shot. Be ambitious. This country rewards big shots. As we say, make audacity the feature not a bug. I talked about this with Ali Afzal on RobotBuildersClub, including what has changed for founders. This part has not changed for me. Comment “session” below if you're building in Physical AI and want a 1-1 call with me. Would love to hear what big problem you are taking a shot at. #PhysicalAI #Founders #Startups
22
There are around 40M existing machines in the heavy industry, and another 6M new machines enter every year. The machine in the heavy industry will not come in gold color. But the underlying hardware architecture, smart software and AI that we are seeing in robotaxis will come to heavy-industry machines. For founders building in Physical AI, I think the opportunity is not only about building the next machine. It is also about how we make millions of existing machines smarter. If you're building in Physical AI, what machine or workflow are you trying to make smarter? Comment below. Would love to hear what you're working on. #PhysicalAI #Robotics #HeavyIndustry #Founders
26
Ask an OpenAI model to make a car fly and it will just do it. It doesn't know a car can't fly. There's no understanding of gravity in there. A physical AI model has the physics coded in. That's the fundamental difference. What's AI still getting wrong about the physical world?
99
In heavy industry, like construction, mining, the big contracts can run into the billions. It's a long game built on trust. You lay out the full vision, a ten year partnership worth hundreds of millions or more, then you find the smallest first piece and sign that. Deliver amazing results, earn the trust, and keep going. That small win becomes a bigger one, and eventually you're closing deals you never could have landed at the start. Comment "session" below if you're building in Physical AI and want a 1-1 call with me.
1
101
Ever think about how all work basically splits into two buckets? There's digital work, the talking, writing, building reports, the stuff AI agents are starting to handle. And then there's physical work, where we use our hands. Moving things, making coffee, carrying logs, building. Take the whole global GDP and every bit of it falls into one of those two. That second bucket is where I'm spending my time. Comment "session" below if you're building in Physical AI and want a 1-1 call with me. hashtag#PhysicalAI hashtag#Robotics hashtag#AI Clip below 🎧
1
69
A founder asked me when to start thinking about revenue. Short answer: day one. Unless you're solving something that truly takes years to pay off like biotech or going to Mars, get revenue early and the rest tends to follow. Comment "session" below if you're building in Physical AI and want a 1-1 call with me. #Revenue #Founders #Startups Clip below 🎧
38
I had a great time talking to Ali on his Podcast. He is doing a great job. Here is a fun short clip on fundraising. Let me know your thoughts and comments below. Comment, "session" below if you are interested or building in physicalAI, and want to have a 1-1 call to discuss with me. #Fundraising #Founders #Startups Clip below 🎧
1
41
I used to use @Grammarly all the time. With @GeminiApp integrated into my workspace, I find myself not using @Grammarly for a while.
1
75
The future where Work is optional is a great future. It’s like singing is optional and only the great singer sings. Same way only the amazing great who wants to work and build will do it.
60
The future of work will be beautiful and amazing. With AI taking all the mundane and necessary work, the best of the best will work because they love to work and are great at it. Like in basketball, LeBron James plays for the love of the game, and the rest of us watch and enjoy.
82
You have no or very limited control over the things that you "have to do", and the things you have good control over are not the things you have to do.
1
78
I automated the repeatable parts of GTM. Custom agent on ChatGPT: feed it your product → get a 20-page GTM plan using my frameworks. More on this with @/Mangomagicpod
1
1
134
Is there an AI bubble? The one that thinks there is an AI bubble, it is true for them. Rest 2% who can’t contain the excitement about AI will experience the unbounded growth due to AI. Our job is to make sure we distribute the growth irrespective of who thinks there is an AI bubble 👍
72
Rest of the world airports just need bit of time to catch up @Tesla
73
Now we have to ask AI about Us 😀
87
Our take after listening to @elonmusk on @theallinpod on the topic of how power efficient our brain is. One way to look at the life and all the progress itself as a small side effect and not that critical part of the overall flow. It keep things in prospective also allow humanity to do more with less stress linkedin.com/pulse/life-seri…
103
Bibhrajit Halder retweeted
Karpathy reviews Tesla self-driving
I took delivery of a beautiful new shiny HW4 Tesla Model X today, so I immediately took it out for an FSD test drive, a bit like I used to do almost daily for 5 years. Basically... I'm amazed - it drives really, really well, smooth, confident, noticeably better than what I'm used to on HW3 (my previous car) and eons ahead of the version I remember driving up highway 280 on my first day at Tesla ~9 years ago, where I had to intervene every time the road mildly curved or sloped. (note this is v13, my car hasn't been offered the latest v14 yet) On the highway, I felt like a passenger in some super high tech Maglev train pod - the car is locked in the center of the lane while I'm looking out from Model X's higher vantage point and its panoramic front window, listening to the (incredible) sound system, or chatting with Grok. On city streets, the car casually handled a number of tricky scenarios that I remember losing sleep over just a few years ago. It negotiated incoming cars in tight lanes, it gracefully went around construction and temporarily in-lane stationary cars, it correctly timed tricky left turns with incoming traffic from both sides, it gracefully gave way to the car that went out of order in the 4-way stop sign, it found a way to squeeze into a bumper to bumper traffic to make its turn, it overtook the bus that was loading passengers but still stopped for the stop sign that was blocked by the bus, and at the end of the route it circled around a parking lot, found a spot and... parked. Basically a flawless drive. For context, I'm used to going out for a brief test drive around the neighborhood to return with 20 clips of things that could be improved. It's new for me to do just that and exactly like I used to, but come back with nothing. Perfect drive, no notes. I expect there's still more work for the team in the long march of 9s, but it's just so cool to see that we're beyond finding issues on any individual ~1 hour drive around the neighborhood, you actually have to go to the fleet and mine them. Back then, I processed the incredible promise of vehicle autonomy at scale (in the fully scaleable, vision only, end-to-end Tesla way) only intellectually, but now it is possible to feel it intuitively too if you just go out for a drive. Wait, of course surround video stream at 60Hz processed by a fully dedicated "driving brain" neural net will work, and it will be so much better and safer than a human driver. Did anyone else think otherwise? I also watched @aelluswamy 's new ICCV25 talk last week (nitter.net/aelluswamy/status/1981…) that hints at some of the recent under the hood technical components driving this progress. Sensor streams (videos, maps, kinematics, audio, ...) over long contexts (e.g. ~30 seconds) go into a big neural net, steering/acceleration comes out, optionally with visualization auxiliary data. This is the dream of the complete Software 1.0 -> Software 2.0 re-write that scales fully with data streaming from millions of cars in the fleet and the compute capacity of your chip, not some engineer's clever new DoubleParkedCarHandler C++ abstraction with undefined test-time characteristics of memory and runtime. There's a lot more hints in the video on where things are going with the emerging "robotics+AI at scale stack". World reconstructors, world simulators "dreaming" dynamics, RL, all of these components general, foundational, neural net based, how the car is really just one kind of robot... are people getting this yet? Huge congrats to the team - you're building magic objects of the future, you rock! And I love my car <3.
2,124
2,112
29,902
12,976,218