saying "the race is over" is so anti-signal as a startup, honestly @e2b with such a culture you're ngmi we'll give you a good race no matter how comfortable you feel you guys better not stop running

the answer is mimicry. most VCs and founders are consensus-driven and once someone else figures out an emerging category, they want to have their horses in the race, too. unfortunately, they’re so caught up in trying to catch up that they fail to realize the race is already over. Sandboxes are dead.

Software is done. You won't believe what I've been making.

Today, I am announcing a pivot. @SirDocsAlot is now a sandbox company. We will build sandboxes for schools. So children can play in them. You buy one sandbox, one sandbox is delivered for free to schools who can't afford it.

You have to invent the future. It doesn't just happen.

Replaced all my CI with a 128 core ampere Altra and it’s literally idling most of the time. If you’re a friend and you are strapped for compute DM me and you’ll get access for €10 per month to help cover internet, electricity and depreciation nixbot.passkey.exchange/repo…

Everyone thinks going down to the metal in infra is easy. It’s not. - You need banks to finance it - You need to work with server manufacturers - You need colocation deals - Then you find out that different colo providers have different power supply per rack, so you can’t just fill racks however you want - So you have to mix and match (sometimes leave empty slots because there’s not enough power) And that’s before getting to the compute, which is a completely different problem in itself.

Because it’s the easiest to use. Really your options are cloud hypervisor (crap performance), firecracker (no major features), or qemu (very very heavy) I don’t mind a little bit of innovation here or there, but we have major companies just straight up vibe forking these security boundaries with no idea how to properly make them safe. This is why we did the hard thing and built our own VMM from scratch - no rust-vmm libraries. We did the formal proofs, DST, fuzzing, etc. and modelled threats properly to close off common attack vectors (like the one that triggered CVE-2026-99999) I fear that 2027 will be the year of sandbox breakouts, but to some extent this will be a good thing. It’ll separate the wheat from the chaff real quick.

new blog - what if i didnt have to let my mac mini just sit there on my desk and could instead use it for sandboxes??

100% true lol

VCs have always checked out when founders stop sending investor updates But today I think even more so The opportunity cost of investing time in a startup that can't be the next Cursor is high If you don't even get monthly updates anymore, well, you just write it off Send them. Especially when times are tougher.

Before, I only had to ignore my emails. Now have to ignore the emails and all the agents trying to make me do something about them.

This is utterly off. Ramp was a hot seed stage company out the gate. They raised capital at $40-100M within a few months of incorporation. At $100M, a 1-2% grant is $250-500K / year in preferred-equivalent terms. Early stage startups simply do not grant that. It's off by more than half an order of magnitude. This is why benchmarks for early stage comp -- especially equity comp, when stated in percentages -- are worse than useless. They are actively harmful, and everyone (founders, people ops, candidates) would do better to completely ignore them. Carta and Pave are useful for base comp, but they are doing the world a grave disservice by pretending to "benchmark" equity (at early stage through Series C/D, it's objectively incorrect). Source: I've spent easily 10,000+ hours working on comp spanning founding through Fortune 500 companies, dating back to 2013. I can't wait to publish a new version of my open source comp tool. (It'll be free, just a public benefit.)

i've started tweeting about boats, idk if u've noticed

When markets are growing quickly, moats don't really matter. It doesn't make sense for leaders to invest in new spaces. And tons of whitespace gets created. When markets slow down, most moats can't save you anyway. Don't worry too much about moats.

The first believer in every startup is betting on the jockey, not the horse.

I moved my some GitHub action runners to a new M5 Ultra Mac Studio and off of @useblacksmith. The price of compute for Blacksmith was $400+/month. You can lease a M5 Ultra 96GB for ~$110/month and have compute to spare.

Probs some truth to that because people are dumb and doing this But we’re gonna get CPU crunched even at its most efficient allocation Because most people will not be running the “efficiency OS”