Agents at scale. One tool to run and manage up to 100,000 AI agents executing millions of tool calls, fast and secure, no matter the scale.
Built for any task, at any scale
Up to 100,000 agents
Orchestrate one agent or a fleet of up to 100,000, executing millions of tool calls in parallel to get work done.
Fast and secure at any scale
Built to stay fast and secure whether you run a single task or saturate a fleet, with permissioned tools and auditable runs.
For everyone, not just engineers
Developers, business operators, and non-technical teams can all put agents to work on the tasks they own.
Any task
Code, research, operations, data work, browser automation: if it can be done with tools, a fleet of agents can do it.
Bring your own LLM
Works with third-party LLMs alongside the Marcia Code models, so you are never locked to a single provider.
On-prem or in the cloud
Deploy where your data and policies require: on your own infrastructure or in the cloud.
Connects your whole stack
Bring everything you already run, new systems and legacy alike, into one agent layer: your tools, APIs, data, and services.
Self-heals and learns
Agents recover from failures on their own and keep getting better at your tasks over time, so the system improves as you use it.
One place to manage it all
A self-contained binary brings the agent runtime, parallel sub-agent fan-out, and a built-in dashboard together: launch agents, watch their tool calls in real time, and steer or stop them, from the command line or the local board.How it fits
Marcia CLI is the command-line surface of the Marcia platform. The same models and account work across the CLI and the Marcia Code API, and fleets scale up through Marcia Fleet.Request access at marciacli.com. For a walkthrough or partnership questions, get in touch.