La Compañía de Gatos
The runtimes

Five runtimes, run in parallel and matched to the work.

Standardizing on one runtime is a single point of failure. We operate five, route each task to the runtime that fits, and stay resilient when any one underperforms.

OpenClaw

Primary workhorse

Drives most conversational flows and tool use across client deployments. The default runtime for end-to-end agent work.

OpenCode

Code & automation

Handles code generation, refactors, and developer-facing automation where correctness and structure matter.

ClawCode

Hybrid runtime

Bridges conversational and coding workloads, used when a task needs both natural dialogue and real code output.

Hermes

Messaging backbone

Powers async messaging pipelines and webhook-driven integrations across every channel we ship to.

Pi

Lightweight agent

Fast, focused agent for narrow, high-frequency tasks where latency and cost are the priority.

Why multiple runtimes?

Fit

Each runtime has strengths — conversation, code, latency. We match the runtime to the task instead of forcing one model everywhere.

Resilience

If a runtime degrades or rate-limits, work fails over to another. No single dependency takes the whole system down.

Cost

Lightweight runtimes like Pi handle high-frequency work cheaply; heavier runtimes are reserved for tasks that need them.

Put a company of agents to work.

Tell us the workflow you want run autonomously. We'll scope the agents, runtimes, and governance — and operate it for you.