Simulating Retail: AI Shopping Agents, European Expansion, and CES

The first community update of 2026 kicks off with our sights set on major deployments. Nils is hitting the road for CES in Las Vegas, followed by meetings in San Francisco and Nashville, as we push our spatial AI technology further into the hands of the world's largest retailers and robotics operators.

Here is a look at what we are working on behind the scenes to make Cactus the ultimate AI copilot for physical work.

Hitting the Road: CES and European Expansion

Momentum continues to build globally. While Nils heads to CES to meet with partners and customers, we are simultaneously finalizing the paperwork on a massive European deployment.

• The European Deal: A minor paperwork hiccup delayed our target December 29th signing, but clarifying meetings are happening this week. Once finalized, this deal will significantly accelerate our footprint across Europe.
• The US Roadshow: Nils will be stationed primarily in the US (specifically San Francisco) through mid-March to drive BD and ensure our earliest robot pilots hit the ground running.

Proactive AI: The "Dungeon Master" Store Simulator

Our goal for Cactus in 2026 is to move beyond passive observation. We want the AI to proactively tell store managers how to run rigorous experiments—like moving products around a shelf—to drive revenue and customer satisfaction.

To build this safely, we aren't just writing code; we are running highly sophisticated simulations inside the Level 10 lab.

• AI Shopping Agents: We are building simulated customers that “shop” in our physical mock store. By using genetic algorithms, we can reverse-engineer shopping personas from real sales data. This allows us to ask the simulation predictive questions like, "Would we sell more milk if we gave it more shelf space?"
• The "Dungeon Master" Test: To truly stress-test our retail copilots, we have appointed an office "Dungeon Master." The simulation will privately alert the Dungeon Master to remove all the oatmeal from a shelf without telling the rest of the team. We then measure exactly how long it takes for the system (and the humans) to notice the out-of-stock event.

As Nils noted: "We want to set up our store from scratch every single week just as an exercise. We want to be looking at the heat maps. We want to be seeing where there might be opportunities to improve sales and run experiments and see if they work."

VLMs for Empty Shelf Detection

We are actively experimenting with replacing traditional, purpose-built empty shelf detectors with Vision-Language Models (VLMs).

Early tests are highly promising. Rather than relying on rigid rules, the VLM can analyze a shelf and reliably identify which specific sections are empty. By tying this directly into store fixture data, we can eliminate the need for brittle, traditional image-matching techniques, making deployments faster and much more resilient to lighting or packaging changes.

Level 10 Expansion and Industry Noise

Our physical footprint in Hong Kong continues to attract top builders in the physical AI space.

• A New Partner Moves In: A new, currently undisclosed partner has officially moved into the Level 10 lab with us. We will reveal who they are as soon as we have a joint demo ready to showcase.
• Addressing the Gossip: We took a moment to address some recent Silicon Valley noise regarding Mentra (the smart glasses company founded by former Auki contributors). Employees from a competing company, Pickle, have been spreading rumors that Mentra is simply white-labeling Chinese hardware. As anyone who has seen the tech knows, this couldn't be further from the truth. In this industry, projection is a powerful thing.

2026 is about execution, scaling our network, and putting physical AI to work.

If you want to join the off-the-record AMA sessions that happen after every stream, make sure to come hang out with us on Discord.

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