Join the movement and participate on Uniswap.
Participate
We had more underway this week than we can put on air. What we can share is concrete: a QR scanner that runs on the phones retailers already carry, an American robot navigating from a real world web map, a gas-sensor inspection PoC on a robot, and the first Cactus release that closes the full capture–analyze–decide–execute loop.
QR detection is how we anchor devices and robots in space. For a long time that path was a GPU scanner tied to Unity — fine in the lab, wrong for the floor.
Retail handsets like Zebra and Bluebird are rugged, often with built-in barcode hardware, and weak GPUs. They will not host Unity cleanly. At demo day we showed a CPU-based scanner outside Unity, plus simulation tooling that loads recordings and synthetic scenes so we can see, frame by frame, where the pipeline works and where it stalls — corners found, identity not yet resolved, and so on.
The goal is reliable recognition on the hardware already walking the aisle.
We also showed BracketBot, an American robot manufacturer, integrated with Auki. After a QR fix it loaded spatial context from the real world web and navigated inside the domain. The white marker on screen was the robot’s own pose estimate — not a separate viewer painting a pin.
Navigation still has work left. Dense spatial semantics should stop it cutting through thin parts of a point cloud. The milestone is simpler: the machine knows where it is relative to our map. Most of the team saw that end-to-end for the first time this week.
On the commercial side, we have a live lead to deploy this class of robot in Asia — a prospective customer talking about ~800 units at around $1,000 per robot per month. That is pipeline shape, not a signed fleet.
A new partner builds sensors for leaks such as CO₂. Their baseline is still manual: people carry the sensor site to site. We finished a PoC with the sensor on a robot. It worked well enough that an existing retail client is opening after-hours store access for real trials.
Product framing is an optional package: pay for the gas sensor add-on and the robot runs autonomous compliance checks. Retail is the shared pipeline; industrial and hazardous sites are the natural extension.
We properly launched the first proactive Cactus — a store brain that tells you how to run the store better, not another dashboard preview.
Last week we framed physical operations as the CADE loop: capture, analyze, decide, execute. This release is the first Cactus cut meant to touch the whole loop. AI recommendations surface how to operate better; approve them and they land in the AR task manager for people to carry out.
That is the product thesis made literal: close the loop, don’t just report on it.
Demo day ended with something none of us had on the agenda. A Cactus client-success lead had built a full arcade-style onboarding game — order picking without Cactus (wander, dodge shoppers, no route), then unlock Cactus (the gag economy prices it at ~$500/month) and replay with navigation and route optimization. Later levels cover domain setup, product scanning, planograms, and a boss fight against Legacy Systemius (spreadsheets).
Half training material, half sales story — continuous with the same person’s earlier IKEA-style setup manuals. We’re looking at hosting it so you can play.
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
