How it works
At a high level, a task moves through the same general path every time — the specific resource it lands on depends on what the task needs and what's available.
You submit a task
Through the web app, desktop client, mobile client, or API/CLI — one place to describe what you need done.
GemmaNode analyzes the task
The request is reviewed at a high level to understand its general nature — coding, research, generation, analysis, and so on.
Available resources are considered
GemmaNode looks at what's currently available across the resource pool — local hardware, cloud compute, AI providers, and agent frameworks.
Work is assigned to a suitable resource
Based on general factors like capability, availability, and cost. The exact selection logic is internal — this page describes what happens, not how it's computed.
Independent work can run concurrently
When resources allow, multiple agents or parts of a task can execute in parallel rather than strictly one at a time.
Results come back to you
Output is collected, consolidated, and returned through whichever client you started from.