Most roboticists think people love robots — because roboticists love robots. That fundamental misunderstanding, according to this conversation, is holding back the entire industry from mainstream adoption.

The discussion unpacks a critical gap in the robotics stack that has nothing to do with dexterity or computer vision. While foundation models and scaling laws are unlocking short-horizon skills — teaching robots to pick up objects or navigate obstacles — they don't solve the problem of long-horizon autonomy. Industrial inspection robots today require manual mission planning: pointing and clicking on maps, scripting task lists that break the moment the environment changes. The challenge isn't making robots more capable; it's making them autonomous enough to operate unsupervised for hours, days, or weeks — reacting to battery levels, rerouting around obstacles, and prioritizing tasks without constant human oversight. This is where decision intelligence comes in: translating strategic goals into tactical actions, enabling robots to think not just about the next five minutes, but the next five days. The conversation also explores why robotics can't simply copy the language model playbook, the third wave of mobile robots in unstructured environments, and why the ultimate goal is to make the robot disappear entirely.

🤖 Why robotics can't rely on the same scaling laws as language models — data scarcity, reliability requirements, and on-device compute constraints make the problem fundamentally different

🧠 The distinction between skills (short-horizon control) and strategy (long-horizon decision-making) — and why both are needed for true autonomy

🏭 How unstructured environments — oil rigs, construction sites, warehouses with humans — demand a new approach to robot tasking

⚡ The productivity case for robots in aging economies with stagnant growth and onshoring pressures

🎯 Why the end goal is making people forget the robot exists — focusing on outcomes, not the machine

Subscribe to Aulium for more conversations on innovation and private capital.

About Kirsty Lloyd-Jukes [CEO of Stateful Robotics]:Co-founder and CEO of Stateful Robotics, a decision intelligence software company spun out of the University of Oxford. Previously ran investor relations, fundraising and corporate development at Waymo for over four years, including the company's first external round of over $3bn in 2020 and a $2.5bn Series B in 2021, and before that sold an Oxford spinout to Waymo and spent time in consulting at Oliver Wyman.About Aulium:Aulium is a video show exploring Innovation and Private Capital, hosted by Thomas Viguier.Connect with us:➡ Aulium on LinkedIn: https://www.linkedin.com/company/aulium➡ Thomas Viguier: https://www.linkedin.com/in/tviguier/