Agent Usage Guide

This guide explains how to automate work and improve efficiency using agents provided by AIP.

Agents are divided into organization agents and personal agents.

• Organization agents: Agents created by administrators and shared with organization users.
• Personal agents: Agents that users create and configure for their own work style.

Common Features

• View and select available agents
• Automate work through chat with agents
• View agent details
• Perform tasks using MCP Integrations and Skills
• View agent execution history and results

Using Agents

Viewing the Agent List

Click Agents in the left menu to view the agents you can use.

On the Agents screen, the following tabs may be available depending on agent type.

• Personal Agents: Agents that you created yourself
• Organization Agents: Agents shared by the organization

Agent Chat Features

• You can visually check context window usage at the bottom right of the chat input field. This helps you identify when to start a new conversation before the agent’s space for processing conversation context becomes full and response quality degrades.
• While the agent is generating a response, progress status text and elapsed time are displayed together, so you can intuitively understand the response process.
• The footer of each agent response shows the credits used for that response.
• When tools are called, the MCP Integration name, Tool name, and Skill name are displayed together, so you can clearly see which source the agent is using to perform the task.
• Agents automatically provide a Built-in Skill for document file processing (PDF, Excel, PPTX, DOCX analysis). Uploaded documents can be analyzed quickly and accurately without additional setup. Files are supported up to 200 MB per turn. There is no cumulative limit per session.
• AI Image Generation: Agents connected to an image generation MCP can generate images directly from text prompts. Describe the image you want in the chat, and the agent generates it so you can view it directly in the conversation.
• Web app and widget creation: In agent chat, you can ask the agent to create a Web app or data visualization widget. Conversation results can be turned into executable forms such as business dashboards, search screens, or simple internal tools.
• TTS responses: You can listen to agent responses by voice. The voice follows the TTS voice setting in User Settings.
• Dedicated chat-only links: You can use links that open the chat screen for a specific agent directly. Users who receive the link can talk to that agent without going through the agent settings screen.

Organization Agents

Organization agents are agents created by administrators and shared with organization users. Users can select and use agents shared by the organization.

Key Features

• Organization administrators configure the agent’s purpose, description, Instruction, Integrations, and Skills.
• Users can select a shared organization agent to start a chat.
• If the administrator configured “Require User Authentication”, users must enter authentication information themselves.

Organization Agent Details

On the organization agent detail screen, you can view the following information.

Overview tab:

• Detailed description of the agent’s functionality
• Information about services that can be connected

Configuration tab:

• Service integration settings that require user authentication
• Suppress personal memory setting
• Dedicated chat-only link for the agent

Suppress Personal Memory:

• If you turn on the Suppress Personal Memory toggle at the bottom of the Configuration tab, the agent will not save the user’s personal memory or use it in future conversations.
• If the administrator disabled personal memory at the agent level, a Locked by Admin badge is displayed and users cannot change the setting themselves.

Dedicated chat-only link:

• If a dedicated link is provided for an agent shared by the organization, users can start chatting with that specific agent directly through the link.
• Link access permissions can be configured together.
• This is useful for workflows such as customer support, internal FAQs, and approval requests where users should be routed directly to a specific agent.

Personal Agents

Personal agents are agents that users create and configure themselves. You can configure the model, instructions, Skills, and Integrations for your personal work style and use the agent directly in chat.

Key Features

• Users can create and delete personal agents themselves.
• Users can import a personal agent based on an existing Preset.
• Users can configure the agent name, description, and icon.
• Users can write the agent description and example prompts themselves.
• Users can configure the LLM model, Instruction, Skills, Integrations, and generation options themselves.

Creating a Personal Agent

In the Personal Agents tab, click the New Agent button to create a personal agent.

Basic information:

• Name: Enter the personal agent name.
• Summary: Enter a short description to display in the list.
• Icon: Upload an icon to distinguish the agent.

After creation is complete, you can configure additional settings on the personal agent detail screen.

Import from Preset

Use the Import from Preset button to create a personal agent based on an existing Preset.

After you select the Preset to import, a personal agent is created based on the settings included in that Preset. Some Integrations may require re-authentication after import.

Personal Agent Details

On the personal agent detail screen, use the Overview and Configuration tabs to manage how the agent works.

Overview tab:

• Description: Describe the agent’s role and how to use it.
• Example prompts: Register example questions that users can refer to when starting with the agent.

Configuration tab:

• LLM Model: Select the model the agent will use.
• Instruction: Write the instructions the agent should follow.
• Skills: Select the Skills the agent will use.
• Integrations: Connect MCP Integrations the agent will use.
• Generation options: Configure file upload support, Artifacts usage, maximum output tokens, and whether to save personal memory.
• Use Web apps, widgets, and TTS: You can ask the agent to create Web apps or widgets and provide responses by voice.

Important Notes

1. Authentication information management: Safely manage information for services that require personal authentication.
2. Permission verification: Regularly review the permissions for data and services that the agent can access.
3. Skill changes: If you modify a Skill, start a new chat so the changes are applied reliably.
4. Sensitive information protection: Handle requests that contain personal or confidential information carefully.