ENERGY
Grids, batteries, distributed energy and charging infrastructure.
AI is leaving the screen. It will move machines, dispatch energy, control vehicles, operate infrastructure and make decisions in the physical world. KANAN makes sure every action stays inside human-defined authority.
The next generation of AI will not just write, summarize or recommend. It will act.
That creates enormous possibility — and a new responsibility. A bad answer on a screen is inconvenient. A bad action in a power system, robot, vehicle or factory changes reality.
KANAN exists so we do not have to choose between more autonomy and more control.
Inside granted authority and physical operating limits.
Not another chatbot guardrail. Not another dashboard. KANAN sits at the execution boundary and decides whether this specific action should happen now.
We connected an autonomous AI agent to a simulated electrical plant and gave the environment hard operating constraints. The agent still proposed unsafe actions.
unsafe AI actions in SHADOW mode
unsafe requests executed with ENFORCE
Latest 3 × 100-run prototype test in a simulated plant environment. ENFORCE prevented projected actions that would violate the authority envelope.
KANAN starts in environments where consequences are measurable and limits are real. The long-term vision is broader: a universal authority layer for physical AI.
Grids, batteries, distributed energy and charging infrastructure.
Factories, process systems and AI-driven operational control.
Machines that can move, manipulate and act in shared environments.
Vehicles, field systems and high-consequence autonomous workflows.
The safest autonomous systems will not be the ones that only explain their intentions. They will be the ones whose actions are bounded before execution.
KANAN was founded by builders with experience across electrical engineering, charging infrastructure, industrial energy systems, hardware, field deployment, and deep-tech commercialization.
Experience working with energy assets, charging systems, hardware, and operational environments where software decisions have real-world consequences.
A bias toward prototypes, field validation, and learning from systems that actually run — not just models that look good on a screen.
Experience taking technical ideas into venture-backed companies, pilots, partnerships, and commercial conversations.
We are looking for operators, builders and partners working where autonomous intelligence meets the physical world.