Navid Aghasadeghi

Co-founder & CEO @Shiraz_AI (YC S26), ex-Boston Dynamics, building robots that learn on the job

[Navid Aghasadeghi @realnavida]

@chris_j_paxton Great blog

[Navid Aghasadeghi @realnavida]

Yup! Looks like we kicked off the ICL frenzy

[Chris Paxton @chris_j_paxton]

In-Context Learning Results Hint at a “GPT Moment” for Robotics

For general-purpose robots to be useful and economical, you need to be able to teach them new skills on the fly. And now we can see how this will work, with in-context long horizon video demonstrations shown by companies like Generalist, Skild, Rhoda, Rhoda, and RobbyAnt.

The idea is simple: prompt with a demonstration, mixed video and proprioception information. Language isn't enough -- it doesn't specify the problem well enough. And it's starting to really work.

more in blog post

[Navid Aghasadeghi @realnavida]

@kushagrchitkar Awesome, congrats!!

[Navid Aghasadeghi @realnavida]

@kshenster Great post!!

[Navid Aghasadeghi @realnavida]

@jonnysalama @ycombinator Super cool!!

[Navid Aghasadeghi @realnavida]

@EzMhypeL @Shiraz_AI Great question! You can think of the example as a prompt that we provide to the robot. So yes a human will then evaluate the robot behavior given the prompt (example), and you can do this instantly becuase we are not retraining / fine-tuning the model.

[Navid Aghasadeghi @realnavida]

Everyone in robotics is waiting for the ChatGPT moment but … missing the most important part of the ChatGPT playbook.

So we left Boston Dynamics and Amazon Robotics to fix that, and start @Shiraz_AI to build robots that learn on the job.

When GPT-3 was released, it couldn’t do every task out of the box. But you could show it a few examples in the prompt, and it adapted on the spot.

We’re bringing that to robotics.

Show our robot a task once, and it adapts on the spot.

That’s exactly what our manufacturing customers need, especially when production lines change every week and traditional automation can’t keep up. And we’re already deploying it with our first customers.

Have a factory that’s difficult to automate? We’d love to talk.

[Navid Aghasadeghi @realnavida]

@_k_sridhar @Shiraz_AI Thanks @_k_sridhar! Yes totally, we think in-context learning will open new markets and will also have important implications on steerability.

[Navid Aghasadeghi @realnavida]

@bdono_ Oh sweet, happy to be the subject of your hot take today! :D

[Navid Aghasadeghi @realnavida]

@yousefhll @ycombinator @nourzahzah @runvendo @garrytan Amazing, congrats!!

[Navid Aghasadeghi @realnavida]

@hamudinaanaa @theworldlabs Yeah I’d be down. I’m traveling until Saturday and then back to the lab. Let’s chat this weekend

[Navid Aghasadeghi @realnavida]

@hamudinaanaa @theworldlabs I would say an easy pipeline that can go from videos of our lab setup, to a simulation that respects the visuals and physics of the workspace.

[Navid Aghasadeghi @realnavida]

@zachdotai Thanks a lot Zach! We’ll get you your robot as long as you promise not to hack it! :D

[Navid Aghasadeghi @realnavida]

@mario_bollini Thanks Mario!! Yes totally, being able to adapt to a new workflow is a necessary condition we see for our customers and for delivering value at scale.

[Navid Aghasadeghi @realnavida]

@broodsugar Nice!! We’ll send our robot to hangout :)

https://t.co/1hf2sRvI5W

[Navid Aghasadeghi @realnavida]

Everyone in robotics is waiting for the ChatGPT moment but … missing the most important part of the ChatGPT playbook.

So we left Boston Dynamics and Amazon Robotics to fix that, and start @Shiraz_AI to build robots that learn on the job.

When GPT-3 was released, it couldn’t do every task out of the box. But you could show it a few examples in the prompt, and it adapted on the spot.

We’re bringing that to robotics.

Show our robot a task once, and it adapts on the spot.

That’s exactly what our manufacturing customers need, especially when production lines change every week and traditional automation can’t keep up. And we’re already deploying it with our first customers.

Have a factory that’s difficult to automate? We’d love to talk.

[Navid Aghasadeghi @realnavida]

@arjunjchaliha Great meeting you guys yesterday!

And here is what I was telling you about :)

https://t.co/1hf2sRvI5W

[Navid Aghasadeghi @realnavida]

Everyone in robotics is waiting for the ChatGPT moment but … missing the most important part of the ChatGPT playbook.

So we left Boston Dynamics and Amazon Robotics to fix that, and start @Shiraz_AI to build robots that learn on the job.

When GPT-3 was released, it couldn’t do every task out of the box. But you could show it a few examples in the prompt, and it adapted on the spot.

We’re bringing that to robotics.

Show our robot a task once, and it adapts on the spot.

That’s exactly what our manufacturing customers need, especially when production lines change every week and traditional automation can’t keep up. And we’re already deploying it with our first customers.

Have a factory that’s difficult to automate? We’d love to talk.

[Navid Aghasadeghi @realnavida]

@DenXX Oooh nice! We’ll test it

https://t.co/1hf2sRvI5W

[Navid Aghasadeghi @realnavida]

Everyone in robotics is waiting for the ChatGPT moment but … missing the most important part of the ChatGPT playbook.

So we left Boston Dynamics and Amazon Robotics to fix that, and start @Shiraz_AI to build robots that learn on the job.

When GPT-3 was released, it couldn’t do every task out of the box. But you could show it a few examples in the prompt, and it adapted on the spot.

We’re bringing that to robotics.

Show our robot a task once, and it adapts on the spot.

That’s exactly what our manufacturing customers need, especially when production lines change every week and traditional automation can’t keep up. And we’re already deploying it with our first customers.

Have a factory that’s difficult to automate? We’d love to talk.

[Navid Aghasadeghi @realnavida]

@lukas_m_ziegler The foundersss 🤔 ;)

https://t.co/1hf2sRvI5W

[Navid Aghasadeghi @realnavida]

Everyone in robotics is waiting for the ChatGPT moment but … missing the most important part of the ChatGPT playbook.

So we left Boston Dynamics and Amazon Robotics to fix that, and start @Shiraz_AI to build robots that learn on the job.

When GPT-3 was released, it couldn’t do every task out of the box. But you could show it a few examples in the prompt, and it adapted on the spot.

We’re bringing that to robotics.

Show our robot a task once, and it adapts on the spot.

That’s exactly what our manufacturing customers need, especially when production lines change every week and traditional automation can’t keep up. And we’re already deploying it with our first customers.

Have a factory that’s difficult to automate? We’d love to talk.

[Navid Aghasadeghi @realnavida]

@liu730chaoqi @RoboticsSciSys @WinstonGu_ @xshenhan @HaonanChen_ @__yuezhao__ @du_yilun @eadeli Amazing!! We should talk!

https://t.co/1hf2sRvI5W

[Navid Aghasadeghi @realnavida]

Everyone in robotics is waiting for the ChatGPT moment but … missing the most important part of the ChatGPT playbook.

So we left Boston Dynamics and Amazon Robotics to fix that, and start @Shiraz_AI to build robots that learn on the job.

When GPT-3 was released, it couldn’t do every task out of the box. But you could show it a few examples in the prompt, and it adapted on the spot.

We’re bringing that to robotics.

Show our robot a task once, and it adapts on the spot.

That’s exactly what our manufacturing customers need, especially when production lines change every week and traditional automation can’t keep up. And we’re already deploying it with our first customers.

Have a factory that’s difficult to automate? We’d love to talk.

[Navid Aghasadeghi @realnavida]

@Jeffjinlin We should talk

https://t.co/0lTpO8lqkL

[Navid Aghasadeghi @realnavida]

Everyone in robotics is waiting for the ChatGPT moment but … missing the most important part of the ChatGPT playbook.

So we left Boston Dynamics and Amazon Robotics to fix that, and start @Shiraz_AI to build robots that learn on the job.

When GPT-3 was released, it couldn’t do every task out of the box. But you could show it a few examples in the prompt, and it adapted on the spot.

We’re bringing that to robotics.

Show our robot a task once, and it adapts on the spot.

That’s exactly what our manufacturing customers need, especially when production lines change every week and traditional automation can’t keep up. And we’re already deploying it with our first customers.

Have a factory that’s difficult to automate? We’d love to talk.

[Navid Aghasadeghi @realnavida]

@_joe_harris_ @SquarePegCap @Tesla @Waymo @AnthropicAI @OpenAI Yes for sure. Are you based in the bay?