When Airports and Robot Giants Knock

Two Surprise Conversations in One Day

This week started with two genuinely unexpected moments in the lab.

First, a delegation from “one of the world’s coolest and most visited airports” just walked into our shared robotics space in Hong Kong. They’re working with our friends at Alpha AI (autonomous drones for inspections), and we ended up giving them a live tour:

• We showed them app-free AR navigation running on our real world web.
• Their reaction: “Could you do this for an airport?”
• Our answer: yes, we absolutely can.

“Let’s bring the world’s coolest AR navigation to the world’s coolest airport. Let’s go.”

The second surprise was even bigger. On a scheduled catch-up with one of the largest robotics companies in China (you’ve heard about them in earlier updates), the tone of the conversation shifted.

They brought in a new senior stakeholder who asked how they could tune an unreleased robot’s features to match our needs. And then dropped: “We want to see about giving you exclusivity for some geographies and industries to protect your interests.”

We didn’t ask for exclusivity; they offered it. That’s a strong signal that they see us as a key piece of their future go-to-market. We’re now reviewing their proposal and planning a visit to their facilities in ~1.5 weeks.

Whether we accept exclusivity or not, the meta-point is clear: we’re now being treated as a strategic partner, not just a software vendor.

Why They Care: The Real World Web

To explain why big robotics players are leaning in, we zoomed back out to the basics:

• 70% of global GDP is still tied to physical locations and labor
• The jump from agentic AI to physical AI (robots, drones, embodied systems) is at least a 3x expansion in TAM
• To unlock that, AI needs access to the physical world

That’s what we’re building: the real world web… an "upside-down internet that lets digital devices browse physical locations."

In short:

• Phones, glasses, and robots connect to Auki domains (our spatial “websites”).
• Our network provides collaborative perception, mapping, and positioning external to any single device.
• Devices can arrive “cold” at a venue, connect to a domain, and instantly navigate and understand the space.

That’s how we pulled off our Terri at WOW Summit demo: we filmed the venue with phones, uploaded that to our network for reconstruction, then had Terri navigate a space he’d never seen before using that shared map.

The Six Layers and Our Order of Attack

We restated our view that general-purpose robots need six layers:

1. Locomotion – moving without falling over
2. Manipulation – interacting with objects
3. Perception – especially spatio-semantic perception
4. Mapping – object permanence and world memory
5. Positioning – knowing where you are in that map
6. Applications – specific tasks and workflows

Most robotics companies start from the top (locomotion, manipulation) and work down. We chose the opposite:

• Focus first on collaborative mapping, positioning, and perception (layers 3–5) as an external service
• Make it so any device—phone, glasses, robot—can tap into it
• Let developers build the applications layer on top

That’s why we say we’re doing app-agnostic, device-agnostic orchestration: phones, robots, and glasses can all share the same sense of space and coordinate around it.

Copilots First, Robots Next

Our strategy stack, in plain language:

1. Build the real world web – the shared spatial network (well underway).
2. Enable the real world app store – let anyone deploy apps for physical spaces (Zappar, Mika, and others are already building).
3. Focus on copilots first – phones and glasses for physical work, where we’re already making real revenue with Cactus.
4. Become a major robot distributor – leverage our existing apps and domains to put robots to work in real venues.
5. Build the OS for embodied AI – once we’re making robot vendors money, we’re in a strong position to define the abstraction layers.
6. Scale to 100 billion devices – the long-term goal: be the default spatial OS for physical AI.

The Chinese robotics partner offering us exclusivity is essentially saying they buy into steps 4 and 5—they’re betting on us to help them sell and deploy robots at scale.

OneShot: A Copilot for Physical Work

To make this concrete, we handed the mic to Mika, a long-time community member and now grant recipient, who’s building OneShot on our network and Mentra glasses.

The origin story is brutal and simple: as an engineering officer on a ship in 2019, Mika forgot to close a valve. Result:

• ~6 metric tons of oil in the bilge
• Two weeks of cleanup
• A lot of damage, plus a permanent nickname: “Bilge Rat”

That experience led him to ask how we can capture and apply operational knowledge better in physical jobs.

OneShot combines three elements:

1. Smart glasses (Mentra) – provide real-world visual context.
2. Checklists – encode institutional knowledge and safe procedures.
3. An AI copilot – watches the task, checks it against the checklist, and guides the worker.

“It’s basically like a superhuman assistant that never gets tired. It’s always available and it has all the knowledge of the job.”

Auki provides:

• On-prem, privacy-preserving collaborative perception (video stays with the customer)
• Spatial context for what the glasses see and where the worker is

And we’re already lining up real prospects for Mika:

• A large Japanese convenience store chain (10,000+ locations, ~8,000 with in-store kitchens) worried about safety, aging staff, and turnover.
• A major Western pizza chain with thousands of locations facing similar problems around mistakes and training.

If OneShot can reduce errors and standardize procedures across these kinds of operations, we’re talking multi-million-dollar annual opportunities just for this one copilot.

Builders Wanted (Seriously)

We closed with a reminder: we can’t build all of this ourselves.

• Almost one third of total AUKI token supply is earmarked for builders on the network.
• We prefer to back 50 serious teams with meaningful grants rather than thousands of tiny ones.
• Typical grants are not “$500 bounties” but tens of thousands of dollars and up—plus:
  • Engineering support
  • Intros to customers and investors
  • On-site time (like flying Mika to Hong Kong)

Promise from Nils: “We’ll fly you to Hong Kong. We’ll give you money. We’ll help you build something cool, and we’ll get you your first couple of thousand customers.”

If you have an IRL use case—whether it’s surgery, warehouses, field service, or something entirely different—and you want to build a copilot or robotics integration on top of the real world web, the best people to talk to are Tracy and Arshak in our Discord.

We’ll be back next week with more on robotics, partnerships, and how these early conversations turn into deployments.

‍

Watch the whole update on X.

アウキ・ラボについて

Aukiはポーズメッシュという地球上、そしてその先の1000億の人々、デバイス、AIのための分散型機械認識ネットワークを構築しています。ポーズメッシュは、機械やAIが物理的世界を理解するために使用可能な、外部的かつ協調的な空間感覚です。

私たちの使命は、人々の相互認知能力、つまり私たちが互いに、そしてAIとともに考え、経験し、問題を解決する能力を向上させることです。人間の能力を拡大させる最も良い方法は、他者と協力することです。私たちは、意識を拡張するテクノロジーを構築し、コミュニケーションの摩擦を減らし、心の橋渡しをします。

ポーズメッシュについて

ポーズメッシュは、分散型で、ブロックチェーンベースの空間コンピューティングネットワークを動かすオープンソースのプロトコルです。

ポーズメッシュは、空間コンピューティングが協調的でプライバシーを保護する未来をもたらすよう設計されています。いかなる組織の監視能力も制限し、空間のプライベートな地図の自己所有権を奨励します。

分散化はまた、特に低レイテンシが重要な共同ARセッションにおいて、競争優位性を有します。ポスメッシュは分散化運動の次のステップであり、成長するテック大手のパワーに対抗するものです。

アウキ・ラボはポスメッシュにより、ポーズメッシュのソフトウェア・インフラの開発を託されました。