frontrun.vc
frontrun.vc
guide

How to Find Stealth Robotics Startups Before They Launch

September 17, 2026·guide·← all articles

robotics stays in stealth longer than software, and for a boring reason: there is nothing to show until the hardware works.

a software founder can put up a landing page and a waitlist in a weekend, so the public trail starts early. a robotics founder may have no shareable artifact for a year. no demo, no screenshot, no beta. by the time the company is searchable, the pre-seed is usually done.

which is the whole opportunity. the company is invisible for a year. the founder is not.

the two receipts

munari: a company before it had a handle

munari builds operator scoring for robot training data. here is the trail, and the dates are public:

  • jul 16 a tracked investor follows the founder. flagged.
  • aug 11 flagged as building.
  • aug 19 website found. still no X account.
  • later the @munariai account goes live, at 6 followers.

47 days between the first signal and the X account existing. during that window there was no company name to search, no handle to follow, and nothing for a database to index. the only thing that existed was a person and an investor who had noticed him.

ixian robotics: a company before it had a name

ixian robotics builds intelligent robots to reindustrialise america.

  • aug 8 the account flagged.
  • aug 18 founder jacob zietek flagged building in stealth.
  • aug 19 published first on our stealth lineup.
  • 21 days later ixian robotics announces a pre-seed from a16z and accel.

the founder flag predates the funding announcement by three weeks, and it predates the company having a public identity at all.

why robotics is the easiest lane to do this in

three things make robotics founders unusually findable early:

the talent pool is small and legible. people leave a short list of places: tesla autopilot and optimus, waymo, google x and everyday robots, boston dynamics, nvidia, anduril, figure, a handful of university labs. a departure from any of them is a meaningful event on its own, in a way that leaving a generic software job is not.

the credentials are checkable. a robotics founder's prior work is usually documented in papers, patents or a named programme. you can tell in ten minutes whether someone can actually build what they are claiming.

investors move earlier. in a category where the product is a year away, investors have to underwrite the person. so the investor follow lands much earlier relative to the round than it does in software, which is exactly the window this signal lives in.

what to actually watch

in rough order of how much they are worth:

  1. a departure from a named autonomy or robotics team with nothing replacing it
  2. investor attention on that person before there is a company. this is the one nobody else can backfill
  3. a bio change to building or founder language
  4. a domain registration matching the founder, often with mail configured before any content exists
  5. a founding-engineer job post for an unnamed company. nobody hires a founding engineer for a side project
  6. patents and papers that stop and start, which for hardware founders is a reliable transition marker

one of these is noise. three that agree, on a person whose career makes the claim credible, is a company.

the honest limitation

this finds the founder, not the thesis. we can tell you a credible robotics person started building and that investors noticed, weeks or months before the announcement. we cannot tell you whether the robot works. that part is still your job, and it always will be.

what the date does is buy you the time to do it first.

keep going

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 →or see trending free

open source deal flow

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

Keep reading