
@SereactAIi
iAccount based inCyprus!
About this account
- Account based in
- Cyprus
- Connected via
- Web
! X says this location may be affected by a proxy or VPN.
Account-level information from X, not a live location or the device used for a specific post.
Embodied AI for robotics
Stuttgart, Germany
Joined July 2023
- Tweets343
- Following110
- Followers1.2K
- Likes91
Pinned Tweet
Sereact has raised $110M in Series B funding. Cortex 2 is live. We're opening our first US office in Boston.
200+ systems. 1B+ real picks. 1 intervention per 53,000.
sereact.ai/posts/series-b
The robot messed up the fold. Then folded it correctly. No one taught it that. Check out the new research from Sereact.
Full house at @NVIDIAGTC Physical AI & Robotics Partner Day. Great way to kick off the week in Berlin. 🦾
#NVIDIAGTC #GTCBerlin #PhysicalAI
sereact retweeted
→ @BerkshireGrey — AI + robotics solutions automating pick, pack, and sort operations for retail, e-commerce, and logistics (now owned by SoftBank).
→ @BoostRobotics — Autonomous systems that monitor and maintain data centers.
→ @tennibot — The world's first robotic tennis ball collector using computer vision and AI.
→ @Teradyneinc Robotics — Publicly traded parent of Universal Robots (cobots) and MiR (AMRs); its robotics division is run partly from Massachusetts.
→ @AcuminoAI — Next-generation, AI-powered robot workers that handle dexterous, repetitive, and dangerous tasks, enhancing productivity and safety.
→ @TheHumanoidAI — Building reliable, safe, and helpful general-purpose humanoid robots focused on labor-intensive operations that current automation can't address.
→ Urbx — Third-party logistics company specializing in highly flexible order fulfillment and last-mile delivery for e-commerce.
→ @SereactAI — Physical AI for warehouses and manufacturing; its Cortex "brain" powers robots for autonomous picking, packing, sorting, and returns handling across single-arm cells, dual-arm stations, and humanoid platforms.
The real benchmark for physical AI is production.
300+ live systems across Europe and the US. One model learning from all of them.
Proud to be #5 in @Siftedeu’s AI 100. Thank you to everyone building this with us. 🦾
#SiftedAI100
#PhysicalAI belongs in the real world, not just in the lab.
Next week, we’re showing you exactly how to make that possible.
See our live demo, learn how customers are deploying #robots in their warehouses, and meet the team making it happen.
DM to set up a time.
@NVIDIAGTC
sereact retweeted
Europe’s next era of robotics is taking shape, where automation meets autonomy.
Join us at #NVIDIAGTC Berlin to hear from robotics leaders at Agile Robots, Dexory, @TheHumanoidAI, @SereactAI, and @Wandelbots for a conversation about how physical AI is advancing Europe’s next era of industrial robotics.
Discover how agentic AI, simulation, digital twins, and robot-agnostic software are transforming fixed automation into intelligent systems that can adapt, learn, and operate at scale.
🕒 October 22 @ 10 AM CEST
📅 Add to calendar: nvda.ws/4dbBVlh
Anyone can film one good pick.
The real benchmark for physical AI is production.
Orders that ship on time, again and again.
sereact retweeted
i always wondered how returns are handled. cool use case of physical ai
Every returned package is a mystery box. Nobody knows what's inside.
Our robots handle it anyway.
Today, with @Zalando and @cevalogistics, we're launching AI-powered returns handling in live operations in Germany and Poland.
Two countries live. Built to scale across Europe.
Every returned package is a mystery box. Nobody knows what's inside.
Our robots handle it anyway.
Today, with @Zalando and @cevalogistics, we're launching AI-powered returns handling in live operations in Germany and Poland.
Two countries live. Built to scale across Europe.
Read more here: sereact.ai/de/posts/zalando-…
Generalization is not one dataset.
Adding more data from the same distribution just reinforces blind spots.
Generalization emerges from expanding into adjacent tasks, environments, and interaction regimes.
That’s what actually scales.
Robots don’t fail at the task.
They fail at recovery.
Cortex 1.5 introduces interactive policy patching:
short human corrections + online RL
→ robots improve after deployment.
No retraining. No rules.
Just better behavior in production.
🔗 sereact.ai/posts/cortex-1-5
We analysed ~66,000 bin images from a live system.
85.8% of the volume was empty.
That’s ~€124k per month in opportunity cost.
Not because AutoStore is inefficient.
Because the system is blind.
If you can’t see inside the grid, you’re guessing.
Monta’s new Bleskensgraaf site is the largest AutoStore in the Netherlands.
It was engineered from day one for AI robotic piece picking with Sereact and Reesink.
No retrofit. No SKU teaching.
Just immediate deployment and predictable throughput.
Support stats nobody expects from a robotics company:
4 min response time
13 min average
22 min to fix a critical issue
And the wild part:
Remote support every 53,000 picks
On-site support every 1.8 million picks
Robots are fast.
Our support’s faster.
300+ PPH in production.
600+ with no bottlenecks.
AI handles 70–75% of 500k SKUs.
Active Ants spent 5 years trying to automate picking.
Nothing survived their SKU chaos.
Until this year.
Not a new arm.
Not a new gripper.
Software finally caught up.
Here’s what changed.











