AI-native, end to end
The CLI and Meshware MCP surface every facet below to an AI assistant — enroll a station, ship a release, triage an incident, pull a remote screenshot — through natural language, not four different screens.
Fleet management
A first-class device model, not a bolted-on inventory table. Enroll, group, tag, and place every station — config inherits cleanly from tenant to group to site.
Over-the-air updates
Publish releases and roll workloads out to your fleet on your schedule, with health-gated rollback if something goes wrong.
Health & incidents
First-class, cause-granular health modeling — not a flat alert feed. Incidents key to the exact failing node in the health graph, even when multiple things break at once.
Remote diagnostics
Pull a screenshot, open a remote shell, or read the logs on a device without a truck roll.
Ops, with an AI copilot
Meshbot
Your fleet, in plain English
Ask a question, get an answer. Meshbot sits in every dashboard and every
incident thread (@meshbot) — a read-only assistant that
already knows your stations, your incidents, and your config. It runs on
your own model provider, so your fleet data never leaves your control.
Meshware MCP
Bring your own agent
Meshware speaks Model Context Protocol natively. Point any MCP-capable agent at your fleet and it reasons over a typed tool catalog built for agents — not scraped CLI output, not a raw terminal. Your agent, your workflow, our fleet data.