Sereact, a Stuttgart based robotics startup, closed a $110M Series B for its Cortex 2.0 'robot brain' that combines world models with vision language action (VLA) models for warehouse picking tasks. The company claims 53,000 picks per human intervention—a figure the article found…
Sereact says its warehouse robots can go 53,000 picks before a human has to step in. The Stuttgart-based robotics startup according to The Robot Report has built a "robot brain" called Cortex 2.0 that it claims approaches full autonomy on repetitive warehouse tasks. The company has the numbers: 200 systems deployed, 1 billion picks completed, a $110 million Series B closed Monday.
What it doesn't have is anyone willing to show its work.
The 53,000-picks figure, which Sereact's CEO cited as the core evidence for his product's reliability, has no published methodology, no customer confirmation, and no independent audit. When asked how the number was calculated, the company pointed to internal logs. No customer has publicly verified the claim. Nobody outside the company knows what counts as a "pick," what counts as an "intervention," or whether those logs have been reviewed by anyone with no financial interest in the answer.
Cortex 2.0 combines a world model, which lets robots simulate outcomes before acting, with a vision-language-action (VLA) model, which connects what the robot sees to language commands and physical movement, according to The Robot Report. The combination means a worker can tell a robot to "grab that bin" and the system figures out how, rather than requiring pre-programming for every variation of every task. Sereact's pitch is that every deployed robot generates real-world picking data, which improves the software, which attracts more customers, which generates more data. The compounding advantage that made Google Search dominant and Tesla's autonomous driving hard to copy, applied to warehouse logistics.
European customers include BMW, Mercedes-Benz, and Daimler Truck. The new Boston office will be the company's first U.S. presence. Total funding stands at over $140 million, including a $29 million Series A in January 2025 led by Creandum.
The warehouse robotics market has attracted significant capital. Amazon's robotics program, plus startups like Covariant, have pushed billions into the space. The core problem has always been the same: robots handle predictable tasks well. Anything requiring real-time adaptation, unfamiliar objects, or slight variations in environment has historically required human assistance. Sereact's claims about low intervention rates, if accurate, would represent a meaningful step toward handling that long tail of warehouse exceptions.
The $110 million valuation is partly a bet that the robot brain is as good as the CEO says it is. Without third-party verification of the 53,000-picks figure, that's not a number. It's a claim.