Agents: secured, long-running AI agents that get your tasks done under control
An agent in ZiyaOS is more than a chat window. It has its own account, permissions and budget, works over long periods and checks in when a person has to decide.
Every agent and every workflow has its own account with its own rights and budget, separate from the person using it.
8B
400B
The right model per task
Every agent and every AI step uses the model that fits the task, local ones included. A small one for classifying, for example.
Signals
A new case, an email, a suspicious event: signals wake the right agent.
Crash-safe
Every step is recorded before it happens. After a crash, work resumes where it stopped. Nothing runs twice.
Firewall included
An agent reaches the internet only at targets you have allowed, internal or external.
External agents too
Claude Code, Codex, Cursor and your own agents work through the gateway with the same rules, the same firewall and Code Mode.
Frequently asked questions
What is the difference between an AI workflow, an AI app and an AI agent?
AI workflows, apps and agents differ like this: a workflow is a fixed process that runs on its own. An app is an interface people work with. An agent takes on tasks that need judgment. In ZiyaOS all three use the same connections and rules.
How do you make sure the AI doesn't do anything wrong?
We make sure the AI doesn’t do anything wrong with the central Guard: every action an AI wants to take runs through it. Without a rule or explicit approval, there is no access. Every action is logged: who, when, what, why, from where and to where. The same applies to people, apps and workflows.
Responsibility stays with people: AI output has to be checked by people, and AI permissions and approvals are granted by people. With ZiyaOS you can trace exactly who did what, how and when, based on which approval and from whom.
ZiyaOS counters hallucinations with tools: agents don’t guess, they query data directly from your systems instead of making it up. With Code Mode, a script does the math, not the model. Where it matters, a person approves the result.