Hanming Ye

@theWaddleLabs | math @ harvard | rsi '23

[Hanming Ye @DozenDucc]

Introducing Waddle Labs: Claude Code for robots.

Connect our API to your robot and enter a prompt, then our agents write code to achieve the task in 20 minutes.

@yiding_song @theWaddleLabs

[Hanming Ye @DozenDucc]

@zobotics Do you wanna try out our API and see how it compares to other robot models?

[Hanming Ye @DozenDucc]

@lotusfracture @yiding_song @theWaddleLabs What are RLMs?

[Hanming Ye @DozenDucc]

Introducing Waddle Labs: Claude Code for robots.

Connect our API to your robot and enter a prompt, then our agents write code to achieve the task in 20 minutes.

@yiding_song @theWaddleLabs

[Hanming Ye @DozenDucc]

@clankrmedia @yiding_song @theWaddleLabs Are you talking about https://t.co/ijx9F4bRSL?

[Hanming Ye @DozenDucc]

@clankrmedia @yiding_song @theWaddleLabs LLMs by themselves aren’t good enough (imo the bottleneck is their perception). So the challenge for us is to transfer what they are good at (reasoning and code writing) to robot control somehow.

[Hanming Ye @DozenDucc]

@pushpak_agr @yiding_song @theWaddleLabs Yes monocular depth estimation coming soon!

[Hanming Ye @DozenDucc]

@grx_xce @yiding_song @theWaddleLabs Uk one problem with evals is judging which robot did it better, say at folding shirts. Need a place to get people to rank robot rollouts 🧐

[Hanming Ye @DozenDucc]

@stevencheng @yiding_song @theWaddleLabs Kinematic limits are handled automatically (since we use the robot’s native SDKs). You can also tell the agent about safety limits as a one time setup!

[Hanming Ye @DozenDucc]

@sarthaksrinivas @yiding_song @theWaddleLabs That’s the part we are iterating over the most. It’s a balance between making new skills reusable and not too granular.

[Hanming Ye @DozenDucc]

@alex_brussell @strimblez @yiding_song @theWaddleLabs 👀More evals to run?

[Hanming Ye @DozenDucc]

@strimblez @yiding_song @theWaddleLabs @alex_brussell Ngl your reply speeds are insane. Makes it hard not to use codex

[Hanming Ye @DozenDucc]

@strimblez @yiding_song @theWaddleLabs Thanks Samantha (and congrats!!)

GPT 5.6 and Fable are very close in our evals… Excited to see how they progress

[Hanming Ye @DozenDucc]

@0171ad @yiding_song @theWaddleLabs THANKS ANDY hopefully they work on construction machinery one day

[Hanming Ye @DozenDucc]

@ryanmcharguee @yiding_song @theWaddleLabs Is that something you’d be interested in? Also do you mean open source models you’re running locally?

[Hanming Ye @DozenDucc]

Ankit and YC have made us 1000% more ambitious. They've pushed us to turn our dorm room research project into the tech we have today.

[Ankit Gupta @agupta]

one of the companies i'm working with this batch is launching today with probably the most mind-blowing demo I've seen this year. stay tuned

[Ankit Gupta @agupta]

https://t.co/xps9Dr9qw4

[Hanming Ye @DozenDucc]

@femiositade Thanks Femi! Glad to have met you at the start of my ventures journey and a bit before that

[Hanming Ye @DozenDucc]

@thejonohart @yiding_song @theWaddleLabs We wrote about our approach here: https://t.co/don5iq5Wbu

Curious what you think? You're right eval is a big pain

[Hanming Ye @DozenDucc]

@sincethestudy @amitavkrshna @yiding_song @theWaddleLabs Hi Brian I think Bracket bots are really cool I tried to preorder them ~6 months ago.

[Hanming Ye @DozenDucc]

@hrhraj @yiding_song @theWaddleLabs We prefer to use robots' native SDK for IK! Though for some robots we use our own IK via Pinnochio.

[Hanming Ye @DozenDucc]

@thejonohart @yiding_song @theWaddleLabs Back in Feb we tried using raw CC on a robot. It was... not fun because Claude overthinks. It calculated quaternions for simple motion.

We try to step out of the way of agents as much as possible, while steering them with how we design the tools they can use.