frontrun.vc
frontrun.vc
receipt

We Flagged This Founder in Stealth 21 Days Before a16z and Accel

September 17, 2026ยทreceiptยทโ† all articles

on september 9 2026, jacob zietek published a post on his own site announcing ixian robotics. in his words: "we've closed our pre-seed with backing from a16z & accel."

twenty-one days earlier, on august 19, we had published his name at the top of a public list of founders building in stealth. the company did not have a name yet.

every date below is public and checkable. that is the only reason this is worth writing down.

the timeline

august 8 the company account is flagged. no name, no announcement, nothing to search for. 32 days before the company has a public identity.

august 18 @JacobZietek is flagged building in stealth. at this point the signal is the person, not the company.

august 19 he is published as #1 on our stealth lineup on X. public, timestamped, and 21 days before anything is announced. the entry read:

september 3 33 tracked accounts are following the founder. the convergence is now obvious in the data, and still invisible everywhere else.

september 9 ixian robotics comes out of stealth. pre-seed closed with a16z and accel. the founder's own post confirms it.

a whole company, before it had a name.

why this one is worth checking

most "we called it early" claims are unfalsifiable, because the claim and the evidence are published at the same time. this one is not:

  • the august 19 post naming him is public on X and predates the announcement
  • the september 9 announcement is on his own domain, not ours. in his words on zietek.dev: "we've closed our pre-seed with backing from a16z & accel"
  • the gap between them is 21 days, and neither date can be edited after the fact

that is the whole standard we hold this to. a receipt that can only be verified by us is not a receipt.

the part that generalises

the company was flagged on august 8. the founder was flagged on august 18. but the useful signal was neither of those on its own. it was 33 tracked investors converging on one person who had no company.

an artifact, a registry filing or a domain registration, would have told you an entity existed. it would not have told you that a group of people who are good at this had already decided he was worth watching. that is behaviour, not paperwork, and it is the one signal in this category that cannot be reconstructed afterwards. a follow-graph diff exists only if you were recording on the day it happened.

which is also why it has an expiry. we cannot go back and tell you who was following someone last march. nobody can.

the four other founders on that list

jacob was #1 on the august 19 lineup. the other four, published the same day, all still building:

  • @petecooling formerly visa crypto. bio: "founder/ceo (stealth). reimagining payments infrastructure for global money movement and agentic commerce"
  • @adri_lemercier previously a quant at jump trading, cs and math at stanford, 3x international mathematical olympiad medallist. bio: "building"
  • @robiot early developer at lovable. bio: "19, building"
  • @kyle 2x founder, ceo of clicker and daylight. bio: "exploring what's next"

four of the five bios say some version of "building" and name no company. that is what this category looks like from the outside.

what happens next with this

the timeline above is becoming a product surface, not just a post. every company card on frontrun is getting the dated story behind it: first follow, flagged, founder identified, convergence, announcement.

the reason is that the date is the product. anyone can tell you a company exists. the useful question is when you could have known, and whether that claim survives being checked.

see the signal this article is built on
frontrun tracks the follow graphs of 2,900+ venture investors and flags companies months before rounds are announced.
start a 7-day free trial โ†’or see trending free

open source deal flow

companies and founders surfaced on frontrun. curated, every week.

Keep reading