Traverse (Traverse (YC W26)) is a data research lab solving the hardest problem in AI training: subjective, taste-dependent work. They generate long-horizon training data for frontier models by observing and capturing what specialists actually do on the job. Congrats on the launch, Lance Yan and Zachary Yu!

@traverseso Really smart angle—taste & long-horizon subjective decisions are exactly where synthetic data falls flat. Excited to see how frontier labs use this to level up. Congrats @lanceyyan @thezacharyyu !

@traverseso “I only work on the hardest problems” - @thezacharyyu

@traverseso congrats!!!! @thezacharyyu @lanceyyan

@traverseso @thezacharyyu @lanceyyan Congrats on making it out of Waterloo, excited to see you guys progress :)

@traverseso let’s go @thezacharyyu, incredible stuff!

@traverseso Congrats!!!

@traverseso Huge 🔥🔥

@traverseso Hugeee @thezacharyyu @lanceyyan

@traverseso Observing specialists to train AI. So we're outsourcing the last bastion of human judgment to the thing that hallucinates citations.

@traverseso aura

@traverseso w launch @thezacharyyu @lanceyyan

@traverseso Letss goooo @thezacharyyu

@traverseso congrats!

@traverseso congrats on the launch!

@traverseso The goats

@traverseso So tuff

@traverseso I love seeing this! Next Scale AI!

@traverseso lfg 🚀🚀🚀

@traverseso Let's gooo!

@traverseso subjective, taste-dependent work is exactly the moat that's hard to commoditize. capturing how experts think — not just what they produce — that's the real signal

@traverseso congrats on the launch!!! @lanceyyan @thezacharyyu

@traverseso congrats on the launch! @thezacharyyu

@traverseso LETS GOO!!

@traverseso Congrats 👏

@traverseso congrats! @thezacharyyu @lanceyyan big moves 🚀

@traverseso 🍍 tuff

@traverseso fireeee

@traverseso re:zero call of the witch goes hard

@traverseso Interesting. how does this handle adversarial situations?

@traverseso congrats on the launch @thezacharyyu & @lanceyyan, super tuff

@traverseso This is objectively the weirdest launch video I’ve seen from this batch

@traverseso lfg

@traverseso YES LETSGOO

@traverseso LFG killer team

@traverseso Observing specialists doing real work to train models solves the data quality problem