Join the movement and participate on Uniswap.
Participate
Auki took delivery of a 300-plus kilogram Galbot S1 this week — believed to be the third and fourth units to leave China — while internal demos showed tighter integration between AI anomaly detection and physical robots, and community data produced a clear leap in scanner performance.
The new robot is the strongest semi-humanoid currently on the market. It can lift 20 kilograms per arm and is built for industrial-scale work. Auki is collaborating with Galbot and one additional partner on a scalable retail deployment that targets one of the most painful operational problems in global retail.
Customer expectations have shifted sharply. Two or three years ago the request was “collect more data for our IT teams.” Then it became “help us with the analytics.” Now the ask is simpler: “we just want AI to tell us what to do.”
An external example from one of China’s largest retailers illustrated how far sentiment has moved. That company has already handed decision-making authority to AI agents and recorded over 200% sales growth. The team sees this as validation of the direction they are pushing internally.
One investor — a former enterprise customer now building his own business — described Auki as potentially “the biggest thing to happen to retail since the adoption of barcodes.” The team sees this as bigger than self-checkout or any prior shift.
Yesterday’s internal demo day showed how several previously separate capabilities now work together on the Cactus platform:
In one live example, an item showing unusually low sales turned out to have boxes stacked in front of it. The manager saw the problem on their phone within hours instead of days or weeks.
The core thesis: the biggest bottleneck in retail is not labor cost or availability. It is poor use of existing labor because managers lack visibility into what actually needs doing.
“We’re not replacing the store worker,” the update stated. “We are, if anything, replacing the store manager.”
The weekly community data collection effort delivered immediate gains. After training on images submitted by participants worldwide, the quick-and-dirty ESL and price tag scanner jumped from roughly 40% accuracy to 90%.
Scanning time for a standard fixture also improved dramatically — from 38 minutes to under 20 minutes in a single week. The team is now targeting an ambitious five seconds per electronic shelf label.
The community contribution was explicitly credited for the leap. Submissions were capped at 300 images per person, but the cap can be raised. More data remains the clearest path to further gains on both the scanner and downstream robotics tasks.
Phil, Sam, and the team return to China next week to work with the new “Chonky” Galbot unit in the realistic replica environment. The small team (just over 20 people) continues to push multiple workstreams that must converge over the coming months.
Two of Auki’s oldest customers are now in discussions about investing in the company — a strong signal that the direction is resonating with the people who matter most.
Auki is making the physical world accessible to AI by building the real world web: a way for robots and digital devices like smart glasses and phones to browse, navigate, and search physical locations.
70% of the world economy is still tied to physical locations and labor, so making the physical world accessible to AI represents a 3X increase in the TAM of AI in general. Auki's goal is to become the decentralized nervous system of AI in the physical world, providing collaborative spatial reasoning for the next 100bn devices on Earth and beyond.
X | Discord | LinkedIn | YouTube | Whitepaper | auki.com
