Aumkar

Accelerating robot adoption in critical industries | Prev: @BMW, @Michelin, @UofIllinois, @CarnegieMellon (dropped out)

[Aumkar @aumkar_r]

The recent progress in robotics has been unbelievable. We've seen impressive demos ranging from humanoids to laundry folding robots to robot dogs that can traverse any terrain. However, most of these robots stay stuck in demos. Adoption of robots in factories, warehouses, and other critical industries remains stubbornly slow. Going from a pilot to full-scale robot deployment can take years and is incredibly expensive.

It's not that robots aren't capable of performing these tasks. The real problem is reliability when the stakes are high. In a research lab, failure is just part of the process. In a factory, one mistake can halt production for hours, destroy expensive equipment, or seriously injure someone. That's why companies need robots that work reliably and predictably in all the messy, unpredictable conditions the real world throws at them.

That's exactly the problem we're solving at Parallax. We're building virtual worlds that mirror actual production sites. Robotics companies use Parallax to test, iterate, and train their robots in these worlds. Factories use Parallax to plan layouts for future automation and to validate various robotic solutions to determine if these robots meet the rigorous standards required in production.

If this resonates with you, please drop us a line - whether you want to build this with us or see how Parallax can accelerate your robotics deployment. We're building the future of robotics right now, and we'd love for you to be a part of it!

[Tanmay Agarwal @tanmayxagarwal]

I’m excited to share two key milestones for Parallax Worlds today: a new funding round led by Midas list investors, and an industrial-grade video-to-3D model with an average error of 3mm that needs just an iPhone, all towards a growing vision for robotics simulation that is focused on an abundance of automation in the next decade. We’re hiring for robotics & cloud infrastructure roles, please reach out if our mission resonates! More below & an exclusive from @SiliconANGLE in the thread.

There are two things that state-of-the-art AI can do for robotics: it can allow us to make more intelligent robots, and it can enable us to massively increase the scale & speed of deploying robots in production. Today, I’m grateful to share Parallax’s $4.9M in funding till date towards the latter, led by @pearvc.

Only three metrics really matter when deploying a robot in a facility: cost, throughput, and reliability. Being able to predict, let alone guarantee any of these at an industrial level is a huge challenge for AI-based robots. This creates a conundrum – intelligent robots present the opportunity to automate tasks that could never be automated before and help with near-shoring & labor shortages, yet we will never see them deployed on the factory floor unless we can measure and improve on the above three metrics.

At Parallax, our mission is to help anyone, not just robotics experts, develop, test, and deploy robots at the speed of light. We do this by building photo-realistic simulations of production sites with just an iPhone, and stress-testing robots inside of them to find robot failures before deployment. Our product leverages advanced computer vision methods to perform at industrial requirements – our AI generated models are accurate within 1 cm, with an average error of 3mm. We create meaningful test cases that go beyond naive randomization with AI-agents that run in a full physics-enabled environment.

I’m grateful today to have the support & trust of Mar, legendary investor at Pear & @payamban, a pioneer in the Space industry. I am also excited to welcome to the cap table GS Futures – the VC arm of one of Korea’s largest conglomerates across manufacturing, construction, and retail, @LightscapeVC, @KakaoVentures, @GaingelsVC, Nova Threshold, @Mana_Ventures, and @BoostVC.

Also extremely grateful for the continued support from John & @RTylerCrown at @Unusual_VC who were Day-1 investors in Parallax, @WizLikeWizard at @spacecadet, who has been with us firmly through thick & thin, and @saranormous & @pranavreddy at @conviction (Embed) for helping us maximize our vision for the company. Lastly, a huge shoutout to Monisha & John who were there for our -1 to 0 journey at Stanford, without whom the company may not have even existed.

Proud of our team and all the work we have done. Lots more to do, so now back to work.

[Aumkar @aumkar_r]

Learn to separate outcomes from effort.

It's easy to get trapped in the illusion of progress. You put in more effort, but the outcome doesn't change at the same rate.

The natural reaction is to try to put in more effort, but effort doesn't necessarily correlate to outcomes. The direction the vector is pointing matters far more than its magnitude.

Tying this to RL: you can keep running a badly initialized training for a million iterations and it still wouldn't yield a good result. Putting in more effort helps you learn what doesn't work more often than what works.

This is why continuously updating your internal world model becomes the primary goal. The understanding of the world allows you to know what actions to take and estimate the expected return of each possible action from your current state.

The goal then becomes updating your world model continuously, and taking the action that results in the highest expected return. Create your own world model for your life, figure out what actions produce the best advantage and keep doing those.

Stop measuring effort. Start measuring outcomes.

[Incentivising @incentivising]

https://t.co/zLW6XKmj8g

[Aumkar @aumkar_r]

Was thinking about this a few days ago. I'd referred to it as 'Rome,' though the cathedral is another great analogy. There are so many civilizational problems left to solve that will take longer than a quarter. It just requires practicing delayed gratification and laying the foundation 'brick by brick' to build something that will outlast our lifetimes
https://t.co/sOjmF9s2II

[Aumkar @aumkar_r]

Each brick feels insignificant until you realize that it's not just any building - it’s a monument that'll be remembered forever. Rome wasn't built in a day. Your legacy won't be either

[Aumkar @aumkar_r]

Disagree. The GPT moment wasn't about capability - it was about accessibility. Millions could instantly access AI on a device they already owned.

While robot capabilities are advancing rapidly, we haven't seen a Cambrian explosion in manufacturing yet.

Robots need their *iPhone moment* first. They need to be accessible to anyone and easy to use. We need mass production paired with the technological advances we're seeing with VLAs & RL to get the GPT moment for robotics

[dar @radbackwards]

The GPT moment of robotics is a lot closer than people suspect

[Aumkar @aumkar_r]

Each brick feels insignificant until you realize that it's not just any building - it’s a monument that'll be remembered forever. Rome wasn't built in a day. Your legacy won't be either

[Aumkar @aumkar_r]

Apple introduces ‘Parallax’ to the ‘World’

[Ben Geskin @BenGeskin]

Reframing a photo after it’s been taken with Spatial Reframing ✨ The new iOS 27 Apple Intelligence feature demo (Beta)

#WWDC #iOS27

[Aumkar @aumkar_r]

Been thinking about this today. Everyone is bound by the same constraint - there's only 24 hours in a day. A high agency person wants to make the most of every one, often cramming more things into the day. AI makes doing more easier than ever.

But Steve Jobs' philosophy of 'less is more' has never been more relevant. The true skill isn't doing more. It's knowing which battles are worth fighting.

You can't do everything. So what should you prioritize? The solution is simple. Decide how you want your 'book' to be written. Then work backward to the highest leverage thing you need to do in the 'moment' to get there. And get rid of everything that's not progressing your 'chapter'.

Thanks for this framework @Alfred_Lin

[Alfred Lin @Alfred_Lin]

https://t.co/fnQEM0WuFJ

[Aumkar @aumkar_r]

An 'L' only becomes a Loss if you treat it like one. Treat it like data and it becomes Learning instead

[Race 🕊️ @multiplanet1]

Elon Musk's first wife once described what it's like to watch him fail.

She said he doesn't react the way normal people react. When a rocket explodes, most people in the room go silent. Some cry. Some start calculating the financial damage.

Musk pulls out his phone and starts making calls. Not emotional calls. Engineering calls. "What failed. When can we fix it. When's the next launch." His voice doesn't change. His face doesn't change. The rocket that just cost $60 million is already in the past. The next one is all that exists.

She said it was the most unsettling thing she'd ever witnessed. Not because he was cold. Because he genuinely wasn't affected. The failure didn't register as failure. It registered as data. An experiment that produced results. Results that inform the next experiment.

This is why he wins. Not because he doesn't fail. He fails more spectacularly than anyone in history. He wins because failure occupies zero psychological space. It enters as data and exits as action.

Most people lose not because they fail but because they spend weeks processing the failure before acting again. Musk spends zero seconds. The gap between failure and next attempt is a phone call.

[Aumkar @aumkar_r]

Market update: Pay-per-token is replacing traditional SaaS

[Aumkar @aumkar_r]

https://t.co/NQsA5UNK0d

[Aumkar @aumkar_r]

We're officially outnumbered by robots (at our office)

Looking for talented roboticists to keep them busy. If that's you, DM me!

[Aumkar @aumkar_r]

SpaceX just became the AWS to xAI's Amazon

[Aumkar @aumkar_r]

Amazing article by @oyhsu highlighting some of the largest problems with deploying robots in the real world! If you're excited to turn these impressive robot demos into production-grade systems that don't fail, shoot me a message - we're hiring!

[Oliver Hsu @oyhsu]

https://t.co/OeMgvmFUKg

[Aumkar @aumkar_r]

Learning fast, adapting to feedback, obsessing over a mission, and pursuing it with relentless tenacity will be the timeless formula for building an unassailable moat

[Naval @naval]

Continuous learning is the only defensible moat.

[Aumkar @aumkar_r]

@bipashasen31 @tangiblerobots Congrats on the launch! Looks amazing

[Aumkar @aumkar_r]

Thanks @Scobleizer for covering what we're building at Parallax! If you're a robotics company looking to accelerate your sim setups, please reach out!

We'll also be hosting an Happy Hour in Pittsburg during the Robotics Discovery Day this week. If you're going to be in town, please send me a message, I'll add you to the list. We're hiring for FT roles for Robotics Research and Backend Engineering. We're also hiring interns for next summer.

If you're interested in working with us, shoot me a message

[Robert Scoble @Scobleizer]

The future of simulated worlds with founder of https://t.co/VsxIMEnd0L

Teaches robots how to work on factory floors.

At Llama Lounge, @jowyang’s San Francisco AI event.