> For the complete documentation index, see [llms.txt](https://docs.poweromni.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.poweromni.ai/en/administration-and-configuration/artifacts/agents.md).
# Agents
The Agents menu is where you create, edit, and configure Power Omni AI assistants.
Each agent can have its own name, avatar, description, AI model, reasoning effort, instructions, connectors, and access permissions. This allows you to create specific agents for different areas, such as customer service, finance, HR, legal, technical support, web search, or document analysis.
### Main agents screen
The main screen displays all agents configured in the environment.
On this screen, you can track key information for each agent, such as the model used, budget, amount consumed, and available actions.
This view helps administrators monitor which agents are active, which models are being used, and how much each agent has already consumed.
### How to create or edit an agent
To create a new agent, click Create.
To edit an existing agent, click the edit icon on the row of the desired agent.
When the side form opens, you can configure the agent’s main information.
#### Agent avatar
The avatar is the image that represents the agent within Power Omni.
It helps users visually identify each agent, especially when multiple agents are available in the environment.
Power Omni allows you to configure the avatar in two ways:
With Custom avatar, you can upload your own image for the agent. The image must be exactly 150x150 pixels and must be in PNG or SVG format.
With Library avatar, you can select an image already available in the Power Omni library.
After selecting or uploading an avatar, the platform will display a preview of the image. You can also change or delete the configured avatar.
#### Agent name
The Name field defines how the agent will be displayed to users.
Use simple and clear names that indicate the agent’s purpose.
#### Agent description
The Agent Description field is optional, but recommended.
It helps users understand what the agent is for before starting a conversation.
#### Model
The Model field defines which AI model will be used by the agent.
Different models may have different capabilities, performance, and costs. Choose the model according to the agent’s purpose.
A simple FAQ agent can use a lighter model. An agent that needs to analyze documents, interpret information, or answer more complex questions may use a more advanced model.
#### Default reasoning effort
The Default reasoning effort field defines the level of reasoning the agent should use by default when generating responses.
This setting allows you to adjust whether the agent should respond more directly or think more before answering.
The available options are: None / Low / Medium / High
Use None or Low for simpler agents that need to respond quickly and directly.
Use Medium for most agents, when you need to balance response quality and execution time.
Use High for agents that need to better analyze context, compare information, interpret documents, or answer more complex questions.
The higher the reasoning effort, the deeper the response tends to be. In some cases, this may also impact response time and model consumption.
#### Agent instructions
The Instructions field defines the agent’s behavior.
This is where you define the agent’s role, how it should respond, which tasks it should perform, and which limits it must follow.
Example:
You are an intelligent agent responsible for searching and retrieving user requests from the internet, organizing the results in a clear, objective, and useful way.
Instructions should be clear and direct. The better the instructions, the more likely the agent is to respond as expected.
#### Improve with AI
The Improve with AI button helps adjust the instruction text.
Use this option when you want to turn a simple idea into a more complete and organized instruction.
After using Improve with AI, review the text before saving.
The Expand button opens the instructions field in a larger area.
### Connectors
Connectors allow the agent to access documents, systems, web pages, or external tools.
Examples:
Use document connectors when the agent needs to query internal files.
Use URL Reader when the agent needs to analyze pages provided through a link.
Use Web Search when the agent needs to search for updated information on the internet.
Use connectors such as OpenAPI, Logic App, N8N, Fabric, SharePoint, or MCP when the agent needs to integrate with external systems.
Enable only the connectors required for the agent’s purpose.
### Permissions
Permissions define which users or groups can view and interact with the agent.
This setting helps organize access by area, team, or role.
Examples:
An HR agent can be available only to the Human Resources team.
A finance agent can be available only to users from the finance area.
A general support agent can be available to all users.
{% hint style="info" %} Review the permissions before making the agent available for use.
{% endhint %}
### Best practices
Use clear names for agents.
Add a short description to explain the agent’s purpose.
Configure an avatar to make visual identification easier.
Choose the model according to the complexity of the task.
Adjust the reasoning effort according to the expected response type.
Write objective instructions.
Enable only the necessary connectors.
Monitor the Usage field to understand each agent’s consumption.
Test the agent with real questions before making it available to users.
### Saving the agent
After filling in the information, click Save.
After saving, we recommend testing the agent in the chat to validate its name, avatar, description, model, reasoning effort, instructions, connectors, and permissions.
### Need help?
{% hint style="info" %} If you have questions about creating agents, choosing a model, reasoning effort, instructions, connectors, budget, or consumption, contact Power Omni Support.
{% endhint %}