👀 any strong preferences between the approach in modal 1 vs modal 2 here?

Expanding QSBS is kind of a big deal. Huge win for founders/early employees who hold eligible stock. The partial holding periods (3 yrs - 50%, 4 yrs - 75%) are an even bigger/better change. Unclear to me if this will only apply to new grants or if it applies retroactively.

Do y’all use an off-the-shelf admin dashboard? I couldn’t find one that pulled in everything I cared about and let me wire up custom flows (Supabase + Resend + Stripe etc.). So I had Cursor one-shot a custom one. Super pleased with the result.

It’s easy to fine-tune small models w/ RL to outperform foundation models on vertical tasks. We’re open sourcing Osmosis-Apply-1.7B: a small model that merges code (similar to Cursor’s instant apply) better than foundation models. Links to download and try out the model below!

Founders shape the culture most when they’re not talking. In what they promote. In what they let slide. In what people learn to do without needing permission.

A little guide for early employees to not miss out on millions in tax savings. Tl;dr - push incredibly hard to be able to early exercise your options. - if you still cant early exercise, exercise as options vest! just my $.02 - not financial advice. neilbhammar.com/blog/early-emp…

Holy guacamole the swift dx is rough so far as new learner. And there’s unlimited edge cases to cause failure when actually trying to run any build 😅

Recent models are post-trained for tool-use. At Osmosis (YC W25), we explored how tools effect model performance. More tool training unsurprisingly led to better performance - but there are still lots of 'unforced' errors. Examples and blog link below!

3 GOATS in one launch video, pumped for these 2. Wouldn’t trust anyone else more to make agents smarter! Go Osmosis (YC W25) Kasey Zhang Andy