Agents
What Is an AI Agent in KriyaOS?
In KriyaOS, an agent is the AI assistant you set up for a specific job.
This page explains what that means in plain language before you create, test, or publish one.
Simple Definition
Think of an agent as a reusable AI worker you can shape, test, and later make live.
Common AI Agent Jobs in KriyaOS
That job might be:
- answering common questions
- helping customers
- recommending products or services
- representing a business in chat
What You Are Really Creating
An agent in KriyaOS includes several parts working together:
- Identity: what the agent is called and what role it plays
- Instructions: the rules and guidance that shape how it answers
- Model choice: the AI model behind the responses
- Knowledge: any files or sources you connect so the agent can answer with more context
- Version state: whether you are working on a draft or using a live published version
Why This Matters To A User
This matters because creating an agent is not just naming a bot.
You are deciding:
- what the assistant should help with
- how it should sound
- what information it should rely on
- when it is ready to go live
Draft And Live Versions
KriyaOS treats an agent as something you can improve over time.
That is why you will see two important ideas:
- Draft: the version you are still editing
- Published: the version that is live
This makes it easier to test changes before you show them to customers or teammates.
Where This Shows Up In The Product
Once you create an agent, you will usually move between these parts of KriyaOS:
- the editor, where you shape the draft
- the playground, where you test the current draft
- the published view, where you see what is live
- the integrations area, where you connect published agents to channels or tools
Next Step
If you are ready to make your first one, read How to Create an Agent.