Lapis One is a dedicated Linux computer designed to run AI agents 24/7. It's not a cloud service. It's not a subscription. It's hardware that sits on your desk and keeps your agent session alive indefinitely.
The premise is straightforward: instead of relying on cloud-based agents or browser-based tools, Lapis One gives you a personal computer running Debian 13 Linux that's built from the ground up to handle continuous agent workloads.
What Makes It Different
Most people think of AI agents as cloud-first. You use ChatGPT, Muse, or Claude through an app or web interface. Lapis One flips that: the agent lives on your hardware.
The device comes with Agent KVM, a hardware feature that gives your AI agent keyboard, video, and mouse control of other computers. Connect Lapis One to another PC with a USB-C cable, and your agent can see that computer's screen, move the mouse, and type — all without installing software on the target machine. It's the same thing a person would do sitting at a desk.
The battery keeps the agent's session alive for 2-4 hours if you unplug it, designed as continuity rather than portability. Lose power for five minutes, and your agent picks up where it left off.
Security by Design
Lapis One uses full-disk encryption with LUKS2, with keys sealed in a TPM 2.0 security chip. If the device is lost or stolen, the data stays encrypted and unreadable. The system core runs read-only, so even if someone gains physical access, they can't alter the operating system.
This matters because your agent will have access to other machines and potentially sensitive workflows. Keeping that access controlled to the device itself — not the cloud — means you're not trusting a third-party infrastructure with agent credentials.
The Use Case Question
For most people, cloud agents work fine. Muse handles your scheduling. Claude thinks through your problems. Why run a personal agent computer?
The answer: control and continuity. Lapis One is designed to run workflows 24/7 with minimal setup. If you want an agent that continuously monitors something, responds to changes, or manages repetitive tasks — all on hardware you own — local infrastructure makes sense.
It's the difference between "I use a service when I need it" and "I have a personal system that works for me constantly."
Who This Is For
The device is positioned as accessible to everyone, regardless of technical knowledge. You don't need to be a Linux expert. But the use case skews toward people who:
- Want agent automation without cloud dependencies
- Need hardware-level control of other machines
- Value owning their infrastructure
- Run workflows 24/7
The Larger Pattern
Lapis One represents a shift: from "AI as a service" to "AI as personal infrastructure." It's the same pattern we saw with Claude Code, Cursor agents, and personal AI assistants — the move from chatbots to agentic systems that actually do work.
Whether that work happens in the cloud or on your desk is becoming a choice, not a default.