PHYSICAL AI INFRASTRUCTURE
Facilities that run themselves
Robots, software and intelligence for buildings. Our first robot, Abluo, cleans restrooms on its own and proves every fixture with photos.
- fixtures photo-verified on connected sites, last 90 days
- 4,260
- fixtures photo-verified on connected sites, last 90 days
- of runs finished without human help
- 86%
- of runs finished without human help
- robots deployed in the US, Singapore, Hong Kong
- 20
- robots deployed in the US, Singapore, Hong Kong
- surface disinfection, lab-validated
- 99.99%
- surface disinfection, lab-validated
Working in
- Hospitals
- Malls
- Airports
- Industrial
- Offices
- Schools
Every station, cleaned by the robot
US airport pilot in Florida, September 2026, filmed by the robot’s own camera.
Restroom cleaning doesn’t scale with people alone
Cleaning teams are hard to hire and harder to keep, and nobody can prove a restroom was cleaned properly. Abluo does the repetitive work and logs every fixture, so facility teams supervise instead of scrub and contracts can be priced on verified results. Customers pay back their investment in 14–18 months today, and in under 12 months at US volumes.
Why now: cleaning labour gets scarcer and costlier every year, robot arms cost a fraction of what they did five years ago, and physical AI, vision-language models that guide a robot arm, now lets a robot understand rooms and fixtures it has never seen.
- Wider market: global facilities management spend, 2025 (Business Research Company)
- $1.97T
- Wider market: global facilities management spend, 2025 (Business Research Company)
Hardware, software and intelligence, built as one
We build every layer ourselves, so each one makes the others better.
01 · HARDWARE
Abluo
A purpose-built cleaning robot, designed in Singapore and California: autonomous mobile base, arm, sensors and onboard tools for steam, chemical jet, hot and cold water, blow-dry, and steam and dry vacuum.
02 · SOFTWARE
Hivebotics OS
Fleet monitoring, remote recovery, scheduling and the customer portal, where every run record, photo pair and floor map lives.
03 · INTELLIGENCE
HiveIntelligence™
Our physical AI models: vision and manipulation models that find fixtures and plan each clean. Every run the fleet completes becomes training data for the next.
Proof of clean, fixture by fixture
Every fixture gets a timed before-and-after photo pair and a place on the restroom’s floor map. Customers see their records in a portal, updated daily, which settles disputes and supports performance-based contracts.
Run record
Restroom A · 5 fixtures · 12 min
- Sink S115:06 → 15:071 min
- Sink S215:08 → 15:102 min
- Sink S315:10 → 15:122 min
- Urinal U115:13 → 15:152 min
- Urinal U215:16 → 15:182 min
Sink S2 · photographed by the robot
Before15:08
After15:10
Built to work alongside cleaning teams
Abluo works in live restrooms next to the people who run them. The robot takes the hardest, least-wanted job, and the team does the rest, so one crew can look after more of the building.
- The robot
- Cleans toilets, urinals, sinks and floors, the repetitive work nobody wants, and logs every fixture.
- The team
- Restocks, handles spills and exceptions, and supervises the fleet from the portal.
Built to scale worldwide
Demand is ahead of supply. We sell to facility services providers, airports, hospitals and fitness chains, directly and through distribution partners: Nexus, our national partner covering all 50 US states, and partners in Hong Kong, Taiwan, Europe and Australia.
- units under LOIs secured across the US, Europe and Asia
- 3,000+
- units under LOIs secured across the US, Europe and Asia
- units on order, including US units through our national distribution partner
- 100+
- units on order, including US units through our national distribution partner
- units a month: production ramp, January to March 2027
- 5 → 20
- units a month: production ramp, January to March 2027
Target: 4,000 robots in service by 2030, built on US and Singapore production lines.
Made where we sell
Designed in Singapore and California. US units are made in the US; Singapore-made units serve Asia, Europe and Australia.
Buy or lease
Customers buy outright with a software subscription, or lease through equipment financing so the robot is paid from operating budgets.
Remote-first support
Fleets are monitored and recovered remotely, backed by trained local service partners. On-site visits are about once a month per robot and falling. In the US, sales and service run nationwide through Nexus.
Physical AI for every restroom
HiveIntelligence™ is our physical AI layer. Setting Abluo up in a new restroom takes three to four days today. Our next step, auto-setup, is being built so the robot can map a room it has never seen, find every fixture and plan where to stand on its own, with the aim of bringing setup down to about 30 minutes. It learns from every run record our fleet produces.
Member of NVIDIA Inception
Built with NVIDIA Cosmos
01
Detect
In simulation, it found every sink, urinal and toilet bowl from a text prompt using NVIDIA Cosmos, with no site-specific training.
02
Plan
It then set a head-on cleaning position at a fixed distance for each fixture.
03
Verify
9 of 9 positions placed, with 2.9 cm average lateral error and 0.13° heading error against simulation ground truth. Field trials come next.
Every real clean feeds back in: 4,260 photo-verified fixture cleans on connected sites in the last 90 days.
How the platform works
Restrooms first, then the rest of the building
One platform, the same base, arm, perception, Hivebotics OS and HiveIntelligence, taking on the repetitive work of a building task by task. Restrooms stay the focus until they are won.
Now · In service
Restrooms
20 robots deployed
Hivebotics OS
Run records for every site
Next · Pilots planned
Pest control
Same spray system
Commercial kitchens
Same arm and steam
Rail depots
From 2027
Later · Roadmap
Grounds
New end-effectors
Common areas
Between restroom runs
Whole-site robot teams
Coordinated by HiveIntelligence
The team leading it
Operators from robotics, AI, aerospace, finance and cleaning. Backed by Vertex and Feedback Ventures, with $9M raised.
Rishab Patwari
Co-founder & CEO
NUS Computer Engineering
Nguyen Tuan Dung
Co-founder & CPO
NUS Mechanical Engineering
Vincent Sim
Chief Commercial Officer
Former MD, Kärcher Singapore
Eric See
Chief Financial Officer
Formerly EDB Investments
Kendrick Chin
Head of Engineering
Chartered Engineer, ex-A*STAR
Nicholas Ong
Head of Operations
Formerly Collins Aerospace
Cawley Mah
Head of Sales & Service, Asia
Formerly Kärcher Singapore
Daniel Kung
Head of Legal & Compliance
Formerly IBM, Grab, Citi
Partners
Member of NVIDIA Inception
IN THE PRESS
Most Promising Start-up, Emerging Enterprise Awards 2025
- The Straits Times
- The Business Times
- Ming Pao
- European Cleaning Journal
Start with the hardest job in the building
Facility owners, FM providers and distributors: see Abluo in your building, wherever it is.
Book a demoInvestors: company overview, traction and our private dataroom.
Investor relations