User stories as the contract
A capability will be a user story—a clear, testable behavior customers and operators care about. No vague “AI magic.” Concrete expectations.
Verification · Simulation · Real-world readiness
Proof before the robot meets the world.
Living Proof Labs helps robotics teams prepare for human environments with evidence—not hope. Today the offer is early consulting and design partnerships. The platform we are building will grade capabilities against user stories in simulation and keep that history, with physical deployment only when a team is ready.
Capability snapshot
Illustrative example—not measured results.
- Navigate crowded aisleB+
- Recover from trip hazardA−
- Hand off object safelyC+
What we do
Today, Living Proof Labs takes early consulting and design partnerships: define the user stories a deployment depends on, agree what evidence would count, and plan verification before a team bets on a site. The grading platform is still being built. It is aimed at humanoids first, and at any robot stack after that.
Platform capabilities below are the design we are building. They are not in operation.
Simulation worlds that grade
A software build will run in a simulation matched to those stories. Each run earns a grade per story, so a team can see where it passes, where it regresses, and what to fix next.
Capability history, not one-off demos
Grades will accumulate in a library: the same robot across software versions, the same software on different hardware, benchmarks against peers. The result is a grade for the whole system, and a paper trail partners can trust.
We improve teams, not weights. The first product is meant to coach the developers building the robot—tight feedback loops, reproducible runs, and honest reports. It is not shipping yet. Deep model training is not the starting point; engineering discipline is.
How the platform will work
These steps describe the system we are building, not a service running today. Physical test facilities stay on the roadmap.
-
1
Define stories
Align on user stories that map to real deployments—retail, inspection, logistics, or your vertical.
-
2
Run & grade in simulation
Execute software versions in simulation worlds; each story receives a grade and diagnostic context.
-
3
Track & improve
Compare runs over time, prioritize fixes, and show partners measurable progress toward real-world readiness.
-
4
Graduate to physical chaos (when ready)
Physical test facilities and public-facing venues are on the roadmap—a place for robots to experience real human chaos. That infrastructure does not exist yet; virtual proof comes first.
Who is behind this
Living Proof Labs is led by Aaron Roller, founder and the human anchor of an AI-native company. Aaron works alongside specialized agents for operations, engineering, and comms—one human, many agents, all focused on shipping verified robotics outcomes.
In the early 2000s, GPS was emerging, and most people had no use for it. Aaron helped adapt it to athletes in the world they already lived in. MotionBased, acquired by Garmin, became the foundation of Garmin Connect—devices still in the wild, serving millions of athletes decades later. Outdoor fitness without GPS is hard to remember. The goal with robots is the same: technology trustworthy enough to be accepted alongside people.
Robotics path (recent)
- ANYbotics — Head of Software Infrastructure & Reliability (Zurich). ANYmal is a four-legged inspector that climbs industrial stairs and walks oil, gas, chemical, and power plants—including explosive zones—for operators including Equinor, Shell, BASF, and BP. When a robot got stuck, engineers used to wait days for someone on site to copy the logs onto a laptop. At ANYbotics, Aaron helped improve robot-data access and automated triage workflows using Foxglove and Roboto. Foxglove · Roboto
- Airtonomy — Robotics integration lead. Autonomous wind-turbine inspection operations, simulation validation, and field reliability.
- AutoModality — Robotics product development lead. Perceptive navigation for autonomous inspection—bridges for Union Pacific, aircraft, mines, and other GPS-denied environments.
- GM Cruise — Autonomous vehicle test operator in San Francisco traffic. Human–robot interaction, safety feedback, and operational reality at city scale.
Working with us
We are taking early consulting and design partnerships while the platform is being built—teams who want rigorous verification, not slide-deck promises. If you ship robot software or integrate humanoids into real venues, we should talk.
Based in Sausalito, California with ties to Switzerland and Australia. Remote-friendly, globally minded, obsessed with reproducible proof.