Leveling up: Auki’s November Navigation Upgrade

Last Day in the Lab (For This Year)

This was Nils's last day in the Hong Kong lab for 2025. Next up is a long road:

• San Francisco (Nov 17–24)
• Singapore → Taipei → Shenzhen → Beijing

Multi-Floor & Cross-Domain Navigation Lands

The big technical headline: multi-floor and cross-domain navigation is now wired up in the real world web.

Jason has been deep in the weeds on:

• Representing different floors as separate 2D nav meshes (for robots running Nav2, etc.).
• Creating a “translation key” between floors/domains so a robot or app can reason about how they relate in 3D space.

In the internal demo:

• The blue nav mesh shows the robot’s walkable area.
• It stops abruptly where Jason simulates an elevator or stairwell.
• Off in the distance, you see another nav mesh – technically a different “floor”, but positioned correctly in the same coordinate frame.

The outcome:

You can “sign into” a specific floor, get directions to a point of interest on another floor, and even see through walls to where that POI would be, because the floors are now stitched together.

This matters for malls, airports, and large multi-level venues, not just single-floor grocery stores. The NPM package and code are live and open source; if you build something on top, you can absolutely come to us for a grant.

A Quick 90-Second Auki Pitch (Refresher)

Because there were new faces, we did a speed-run of what we actually do: “Today two-thirds of the world economy is still tied to physical locations and physical labor, out of reach for AI. If we want to go from generative and agentic AI to physical AI and thereby 3x the total addressable market of AI, we must first make the physical world accessible to AI. That’s why we are building the real world web.”

The real world web:

• Makes physical locations browsable, searchable, and navigable for robots and AI.
• Is already rolling out into 1,000+ locations, running AI copilots for physical work on handhelds.
• Is generating millions in revenue today on phones, with glasses and robots coming into the same network.

Nils repeated our near-term ambition: “We are now raising 75 million to put 500 humanoids and 5,000 smart glasses to actual work-work in retail in 2026.”

That’s where a lot of the current BD and fundraising energy is going.

California Roadshow: Airports, Ports, Grocers, Pizza & Airlines

The multi-floor work shipped just in time, because the next couple of weeks are packed with location-heavy prospects:

In California we’re meeting:

• One of Europe’s most trafficked airports
• A major Canadian port
• One of the US grocery retailers with the most robotics adoption today
• One of the world’s largest pizza chains
• One of the big airline alliances

The pattern: large physical venues with complex flows of people and goods, who see value in:

• Navigation (staff and customers)
• Spatial data (what’s happening where, over time)
• Robots that can plug into an existing real world web domain instead of taking weeks to integrate from scratch.

Short trip, dense schedule.

Financing Robots at Scale: GAIB Partnership

We also officially announced our partnership with GAIB, a Web3 finance project focused on tokenizing different kinds of debt and financial instruments.

The goal together:

• Figure out how to finance a large fleet of robots using on-chain structures.
• Let GAIB’s community help fund deployments, backed by real robot productivity and contracts.
• Make it possible to scale to thousands of units without every retailer needing to buy hardware upfront.

We’re still in the unit-economics and structuring phase, but as Nils put it: “The kind of numbers they are throwing around to support this expansion is very promising.”

If you care about the intersection of DeFi and physical robots, this is one to watch.

Retail Rollouts: A Seven-Figure Deal Re-Cut (But Still On Track)

Nils also gave a candid update on the big retail rollout that’s been “days away” for what feels like forever:

• The umbrella retailer has multiple store groups.
• Some of those stores were moving slowly, slowing everything down.
• We restructured the deal to exclude the laggards and move ahead with the decisive part of the group.

Key points:

• It’s still a seven-figure deal.
• We’re now talking directly with the CEO and CFO for the final green light.
• We expect rollout this year.

The encouraging bit: usage. “Since the pilot, we see that they are actually very actively using the software every single day… even more than the Swedes who were our first big enterprise customer.”

They’re also the second live implementation of shopper navigation (the first was Pepito in Bali). Their shoppers are already using AR guidance to find products.

Hiring, Referrals, and BD on the Move

We’re hiring, and this week was heavy on interviews.

• Auki is competitive to get into, so if we pass on you, it’s not a dig – but we really appreciate everyone applying.
• We’re actively growing BD and deployment capacity because the retail pipeline is now measured in 8-figure opportunities.

Johannes (BD lead) joined the call to share:

• He’s heading to North America soon. One of his targets is likely to be our first ≥8-figure account if all goes to plan.
• He’ll also be in Toronto, meeting another large retailer.

We also reminded the community:

• There’s a referral program.
• People who helped with Bali (Pepito) made actual money from deployment work.
• If you know decision-makers in retail or physical venues, introduce them to Johannes.

Under the Hood: Markerless VPS, Motion Nodes, and 15,000 m² in Under an Hour

We closed with some deeper tech updates.

1. Scaling markerless positioning

Our visual positioning system without markers has been pushed beyond the early ~20 m² comfort zone:

• We’re now benchmarking on spaces 10–20x larger.
• Early tests are promising; we’ll publish real numbers once we’ve run enough trials.

This matters for large stores, malls, and airports that don’t want markers everywhere, but still want robust positioning.

2. Motion nodes for real-time tracking

A new node type is coming: the motion node.

• It performs real-time spatial tracking on behalf of glasses or robots.
• Offloading that work to a local node lets devices save battery and run cooler, especially lightweight glasses.
• You’ll be able to experiment with motion nodes early next year.

3. Reconstruction network milestone

Johannes also shared a big win for the reconstruction network:

• We ran 20 reconstruction nodes in parallel.
• They processed a 15,000 m² conference scan (Rethink) in under an hour.
• Global refinement on that size of domain was on the order of minutes.

From Johannes’s perspective as the guy on-site: “The goal there is to pretty much do all the prep work in a grocery store, go for lunch, and come back and the map is done… This brings it down to absolutely one day.”

For a typical grocery store, the full pipeline from scan to usable domain should land in the 5–15 minute range once tuned, which is exactly what we want for real deployments.

We also gave a shout-out to Braintrance, one of our grant recipients, who has been opening a lot of pull requests on the core repos. If you haven’t checked them out yet: braintrance.net.

‍

Watch the whole update on X.

About Auki

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