Join the movement and participate on Uniswap.
Participate
Auki's May 15 community update covered three practical advances: giving glasses real spatial context without sending data offsite, using an AI copilot to speed up physical robot work, and opening the Exocortex's internal tooling to the public.
The lab hosted its monthly Robots & Beers event. Roughly 50 people showed up for robot soccer, presentations, pizza, and beer. Last Friday the team ran the world's first remote first-person view soccer game courtesy of partners at CTRL+R. The format produces a lot of own goals — roughly a third of all scores — but the point is testing real remote operation under social conditions.
The bigger technical demo was remote SLAM. The Mentra Lives glasses have limited onboard compute. They stream video over the local network to a GPU machine that builds the 3D map and tracks position. Feature points are detected and tracked across frames so the system can recover when it loses track and recognize previously seen space.
This matters because glasses without 3D position awareness cannot run useful location-based AI. With a local GPU doing the heavy lifting, the glasses can now provide context to models like Cactus — telling a wearer exactly where they are in a supermarket or office and adjusting navigation if they take a wrong turn. The same coordinate system also lets the system know what the glasses are looking at relative to a digital twin.
The team is heading to Shenzhen next week to meet glasses manufacturers and push for device-agnostic support. The goal is any camera-equipped glasses with an open ecosystem should be able to know their position in the world.
Mika, an Auki grant recipient, is building Oneshot — an AI copilot that runs on the glasses. In a live test, Tracy (never having operated the robot before) used the glasses and Oneshot to unbox and ready Terri. It took under 10 minutes from box to the robot standing and moving on its own. Art, also new to the robot, did the same task the conventional way with laptop, manual, and video and took 25 minutes.
The difference came from the copilot giving step-by-step guidance tied to what the wearer was actually seeing and doing. The team sees this as strategic: they plan to deploy thousands of robots and the glasses are the practical way to handle maintenance at scale without every operator becoming an expert first.
Watch the live Oneshot unboxing test here.
The robots currently rely on off-the-shelf models for detecting price tags, ESLs, and empty shelves. Those models are trained on small, imperfect datasets. The community is being asked to help annotate images from supermarkets, pharmacies, and convenience stores. Contributors earn tokens for the work. The data stays protected and will be used to improve the robots' reliability in real retail environments.
The Exocortex gained a presentation skill that turns a week's GitHub activity and Discord notes into slides in a few minutes. More visibly, the team launched Pulse — a public front end that shows every weekly presentation, a reverse timeline of milestones, repo links, and a "catch up" function. Tell it how long you've been away and it summarizes what happened.
You can also query specific contributors or categories (SDK work, Cactus, network, etc.). An NDA filter keeps sensitive items out. The explicit goal is making "building in public" self-serve so people can check progress without waiting for the weekly call.
The team also completed the ability to speak and hear through the robot over the local network. During Robots & Beers, the robot was used to greet arriving guests and hold conversations. A separate test involved walking the robot across the office to fetch headphones from the podcast studio. Internet access is the next step; the feature should be available for the next demo day.
A hackathon is running this weekend in the Hong Kong lab with participants from across Asia. Demo Day is Sunday. The Shenzhen glasses manufacturer meetings start Monday.
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
