Documentation Index
Fetch the complete documentation index at: https://koreai-v2-home-nav.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
This document covers the key features available to end users in the AI for Work platform.
Overview
| Feature | Purpose |
|---|
| Agents and Flows | Connect with existing agents, create custom agents, and build automated flows |
| Schedulers | Create and manage automated schedules for agents and flows |
| Language Settings | Set a preferred language for the platform interface |
| Enterprise Directory | Search employee contacts, view profiles, and explore the organization chart |
| Projects | Organize files, instructions, and conversations around a specific topic or workflow |
| Uploading Files | Attach files to conversations through the compose bar or drag the file on the compose bar |
| Feedback Submission | Submit feedback on AI responses using upvote or downvote |
| Log in | Access the AI for Work Assistant and begin interacting |
| Agentic Flow | Execute multi-step workflows across connected apps |
Agents and Flows
The Agents and Flows section lets you connect with existing agents available in your workspace, create new custom agents, and build automated flows. Access this section from the homepage by selecting Agents and Flows.
Connecting with Existing Agents
When you open the Agents and Flows section, the platform displays all the agents available to you. These include pre-built agents that administrators have published and shared agents from other workspace members. You can browse, search, and start using any agent directly from this view without additional setup.
To connect with an existing agent, select it from the list. The platform routes your queries to the selected agent based on its configured purpose and capabilities. If the agent requires an external integration (such as Jira or Google Calendar), the platform prompts you to authorize the connection on first use.
Creating an Agent
The agent creation feature provides a streamlined interface for building new agents. The system adapts its behavior based on workspace access and permissions.
Create Agent Button
| Condition | Behavior |
|---|
| User has permissions | The Create Agent button appears in the Agents and Flows section. |
| No workspace access | The system hides the Create Agent button. |
| No agents created | The interface hides filter options (My Agents and Shared Agents). |
- Navigate to the Agent and Flow section and select Manage.
- Select Create Agent to display a list of available agent types.
- Select the desired agent type.
- Choose Create from Scratch or Use Marketplace.
- Select a preferred creation method.
Workspace Routing
| Scenario | Behavior |
|---|
| Single Workspace | The system navigates directly to your Personal Workspace in a new tab with the selected agent type pre-selected. |
| Multiple Workspaces | A workspace selection modal appears, including My Workspace (Personal Workspace) as an option. |
Flows
Flows let you chain multiple agents into a sequential, automated workflow. Each step in a flow uses a specific agent, and the output of one step feeds into the next. Use flows when you need to automate multi-step tasks that span different tools or systems.
You can access flows from the same Agents and Flows section. The platform displays existing flows alongside agents. To create a new flow, select Create Agent and select the Agentic Flow agent type. The creation wizard guides you through defining the flow name, purpose, and individual steps. You can add, edit, reorder, or delete steps before running the flow.
Once you publish a flow, it appears in the Agents and Flows list and you can trigger it from the chat interface or schedule it for automated execution.
Schedulers
The Schedulers feature provides a centralized location to create, manage, and monitor automated schedules for agents, flows, and platform modes.
Accessing Schedulers
- On the homepage, select Agents and Flows.
- Navigate to the Schedulers tab.
The tab displays all schedulers you have created, along with any admin-enforced schedulers assigned to you.
Creating a Scheduler
- On the Schedulers tab, select Create.
- In the Agent field, select an agent you have access to, a flow you have access to, or a mode: Work, Web, or AI Model.
- Set the schedule timing and frequency: Once, Hourly, Daily, Weekly, Monthly, Custom, or Cron.
- In the Trigger Query field, enter the instructions for the agent or flow to execute on each run.
- Select an Execution Mode:
| Execution Mode | Behavior |
|---|
| Ask before running | The platform sends a notification at the scheduled time and runs the agent only after you approve. |
| Run automatically | The platform runs the agent on schedule without approval and notifies you when complete. |
- Select Save.
The platform enables the scheduler by default after saving.
Managing Schedulers
You can toggle any scheduler on or off at any time. Changes take effect immediately.
| Action | Description |
|---|
| Edit a scheduler | Select an existing scheduler to open its configuration. Update the agent or flow selection, schedule timing, or query as needed, then save. Changes take effect immediately. |
| Delete a scheduler | Select the scheduler and select Delete. The platform removes it immediately. |
| Reuse an agent | You can assign the same agent to different schedulers. Each scheduler operates independently with its own timing, frequency, and query. |
Language Settings
Language Settings lets you choose a preferred language for the platform interface. The selected language applies to all static and dynamic UI text across the application.
First Login
On first login, a language selection module appears. Search for or select your preferred language, then select Done. All static and dynamic UI text updates immediately across the application.
Changing Language
You can update your language preference at any time through the profile icon > Language, which reopens the language selection module.
Arabic (RTL)
Selecting Arabic switches the entire interface to a right-to-left layout. All UI elements, including the history panel, response generation area, menus, and icons, realign to RTL. Text input also adapts: Arabic characters append to the right of the cursor and English characters append to the left.
Your individual language preference, once set, takes precedence over any account-level default that an administrator configures.
Enterprise Directory
Enterprise Directory provides centralized employee contact management with intelligent search, customizable filters, and rich contact profiles.
Access the Directory
- Open the application and select Enterprise Directory.
- Contacts appear ranked by designation in a hierarchical view.
- To view your own contact card, go to your Profile section.
Search and Filter Contacts
Use the Search bar to enter keywords for real-time results matching names, designations, departments, or any searchable field. Use Filters to narrow results by Department, Designation, First Name, Location, or any custom filters your administrator configures.
Contact cards display: Name, Designation, Team, Phone, Email, and any administrator-configured fields.
Select a contact card to open a side panel. Expand to full-page view for complete details, including: Profile Picture, Name, Designation, Team, Email, Phone, Location, Time Zone, Employee ID, Manager, Reportees, Team Members, and any custom fields.
When you search for information about other employees in chat, the assistant returns an answer along with a contact card at the end of the response. If you search for multiple employees, the assistant displays multiple contact cards.
Contact cards appear only if the administrator has configured the Employee Directory. If a service account is the lookup source, the assistant maps the email ID to the Employee Directory data and displays contact cards based on the matched employee records.
Agentic mode behavior: If the search fallback is in agentic mode, the assistant passes contact lookup as a tool to the work agent and returns the response within the conversation. If agentic mode is not active, the contact lookup agent responds independently.
Organization Chart
The Enterprise Directory includes an Organization Chart tab that displays your organization’s reporting hierarchy visually. Select the Organization Chart tab on the Enterprise Directory page to load the hierarchy view.
The Organization Chart tab becomes available only after an administrator configures and publishes the Enterprise Directory.
| Control | Action |
|---|
| Zoom In / Zoom Out | Adjusts the scale of the chart |
| Fit Preview | Resizes and centers the chart to fit within the visible area |
| Collapse | Collapses the entire chart to show only the top-level card |
| Full View | Opens the chart in full-screen mode with all interactions available |
Personalisation and Memory
When your administrator enables memory for the account, the AI assistant remembers context from your conversations, references chat history, and draws insights from connected business apps to deliver personalized responses. A one-time notification confirms when memory is active.
Access your personalisation settings from Profile > Personalisation & Memory. The widget contains two tabs: Persona and Memory.
Persona
What Names Do You Go By
Add alias names that your colleagues use to address you at work. You can add up to five aliases and delete any alias at any time. For new users, alias names are blank by default.
Context Gathered About You
The platform generates a summary about you once every 24 hours, covering:
- Personal Context: Residence, interests, preferences, and family information.
- Work Context: Job title, role, team position, manager and reporting structure, key KPIs, projects, and responsibilities.
- Immediate Focus: Summary of recent projects and items you are currently working on.
- Brief History: Recent items (last one to two months), earlier context (three to five months), and long-term work (six or more months).
You can edit the summary by providing instructions. The platform preserves these instructions in Manage Edits and applies them the next time it regenerates the summary. You can also delete your context to reset it; generation restarts from that point.
For new users, the context section displays: “Your context is being generated and it grows as you work. You can also add information yourself to get started faster.”
People You Work With
The platform periodically identifies people you work with from connected business applications and displays them with a colored border. These may change when the platform repopulates the list.
You can also manually add people. Manually added people are never removed and display with a black border. Add internal people by selecting from the directory, or add external people by entering an email address.
Custom Instructions
Add global or agent-level instructions to control how the assistant responds to you. Instructions can cover tone, formatting, intent identification, and response behavior.
Examples: “Whenever sending an email, use my name as Meg”; “Give me answers in exactly 4 bullet points”; “When I want to create a ticket, use project as EVA.”
Add instructions from any of these entry points:
- Profile > Personalisation & Memory > Persona > Instructions > Add Instruction: Toggle on agent-level to scope the instruction to a specific agent. Work, Web, and AI Model are also available as agent options.
- Agents & Flows > Select an agent > Add Instruction: Instructions added here are agent-level by default.
- While chatting: When the orchestrator detects that you shared a reusable instruction, it displays an inline prompt: “This also appears to be a reusable instruction. Would you like me to save it as a preference for future use?” You can edit, save, or discard it inline.
Each agent can have only one instruction. If the platform detects an instruction for an agent that already has one, it suggests an update that appends to the existing instruction.
Memory
Each conversation thread generates a memory entry. The platform captures memory in the following format:
- Conversation Summary: Summary of the thread.
- Key Insights Captured: Important insights from the conversation and what they imply about you.
- Actions & Decisions: Key actions performed and decisions you made.
If a memory instruction was defined during agent creation, the platform considers it when generating thread-level memory for threads where that agent was used.
View and manage memory
- View memory entries in a timeline view.
- Filter memory entries by agent (filter options include all agent names, as well as Work, Web, and AI Model).
- Search across all stored memories.
- Delete individual memory entries. Deleted memories are not referenced in future chats.
- Clear all memory from the overflow menu.
If you delete a conversation thread, the associated memory entry is also deleted.
Referencing past interactions
When you ask questions that reference past interactions (for example, “Based on my previous interactions, where should I go?” or “As I discussed yesterday…”), the platform retrieves the relevant thread from memory and displays it in chat for reference.
Projects
Projects let you organize files, set custom instructions, and maintain focused conversation threads around a specific topic or workflow. Instead of uploading files again or repeating context each time, you create a project once and use it as a persistent workspace for all related queries.
Projects appear in the left navigation panel under Projects. Each project contains its own files, instructions, recent threads, and a dedicated compose bar scoped to the project context.
Creating a Project
You can create a new project from the left panel or the All Projects view.
From the Left Panel
- Hover over Projects in the left navigation panel. A + (Create Project) icon appears.
- Select + to open the project creation dialog.
From the All Projects View
- Select Projects in the left navigation panel to open the All Projects view.
- Select + Create in the upper-right area of the Projects grid.
Project Creation Steps
- In the New Project dialog, enter a Project Name (mandatory) and optional Project Instructions that define how the platform generates responses within this project.
- Select Next to proceed to the file upload step.
- In the Project Files dialog, add files by dragging, browsing your local system, or importing from a URL. Supported file types include PDF, DOCX, PPT, and TXT (up to 25 files, 15 MB each). You can also select Add files later to skip this step.
- Select Done to complete setup. The platform navigates you to the Project Detail page.
Project Detail Page
Opening a project displays its detail view with a two-panel layout.
Left Panel
| Section | Description |
|---|
| Files | Lists all files uploaded to the project. A search bar and a Select all checkbox help you manage files. Hovering over a file reveals an Ask action that selects that specific file as context in the compose bar instead of the entire project. |
| Instructions | Displays project-specific instructions. Select the instructions area to open edit mode and add or modify instructions. The platform applies instructions when you set the entire project or files within the same project as context. If you add files from different projects, instructions don’t apply. |
| Recent Threads | Lists all conversation threads that you started with the project as the active scope. These threads are a filtered view of your main conversation history scoped to the project. If you rename or delete a thread in the main History section, the change reflects here, and vice versa. |
| Section | Description |
|---|
| Smart Suggestions | Auto-generated suggestions based on the project files, displayed at the top of the compose area. Suggestions update automatically whenever you add or sync files. |
| Sample Queries | A section below smart suggestions where you can add or edit custom sample queries to guide project usage. |
| Compose Bar | A project-scoped compose bar preset with the selected project. You can scope down to specific files within the project but cannot deselect the project itself. Only files within the project are available for scoping. The platform generates responses from the uploaded files or from the Work mode orchestrator, with clear citations and sources. |
File-Level Interactions
You can interact with individual files directly from the project detail page.
| Action | Description |
|---|
| Ask on a file | Hover over any file in the left panel to reveal the Ask action. Selecting Ask sets that specific file as the context in the compose bar, allowing you to query against a single file rather than the entire project. |
| Bulk actions | Select multiple files using the checkboxes to perform bulk operations such as deleting, moving, or querying across the selected files. |
Project Threads in History
Project-based threads appear in the History section alongside regular threads, distinguished by a project icon. Selecting a project-based thread opens the associated project, and the selected thread loads within the project with full project context, including instructions and files, preserved.
Sharing a Project
You can share a project with other users by managing its sharing permissions.
To Share a Project
- Open the project options menu (available from the All Projects view or the project detail page).
- In the Share dialog, choose a sharing option:
- Private: Only you can view and access the project. It remains visible only in your Projects list.
- Selective users: Choose specific users to share the project with. They see the project and its files under Shared projects in their Projects list.
- Select Done to apply the sharing settings.
Shared Project Behavior
| Aspect | Behavior |
|---|
| Access level | Shared users receive read-only access. They can’t edit instructions or add, remove, or modify files. |
| Recent Threads | Recent Threads in a shared project reflect only the threads that end users generate when accessing the shared project. |
| Suggestions and queries | Smart suggestions and sample queries that the project owner creates remain visible to shared users. |
Uploading Files
Attach files to your conversation by select the attachment icon in the compose bar or by dragging and dropping a file directly into the chat interface.
Upload by Dragging
Drag a file from your local system over the chat window, on to the highlighted UI area. The progress indicator displays the upload progress for file uploads.
File Storage and Data Handling
When you upload files during chat sessions, the platform can store files in managed cloud storage. The platform protects all files in managed storage with industry-standard encryption to ensure data confidentiality and integrity.
If an administrator disables cloud storage, uploaded files remain available only during the active chat session and the platform doesn’t retain them after the session ends.
Feedback Submission
You can submit feedback on AI responses using the Upvote or Downvote icons in the chat interface. Downvote submissions include a category selection and an option to add comments.
Share Query and Response
If an administrator enables Hide conversation in Data Settings at the agent or account level, the feedback pop-up displays a Share Query and Response checkbox, checked by default. When you enable this option, the platform stores the question and response for that entry in a visible state. If you have prior queries in the same thread, a Share previous queries option also appears, capturing up to the last 5 queries from that thread.
The platform records feedback only when you select Submit. Canceling the pop-up discards the submission entirely.
Log in
Log in to the Platform. The AI for Work Assistant is available on the homepage.
Type an open-ended question or instruction in the Ask or search anything search box. For example, if you connect to Jira, type Get my Jira issues and the Jira agent retrieves your issues automatically.
Agentic Flow
The platform manages complex, multi-step workflows by organizing tasks into a structured sequence, each handled by a specialized agent. Each agent performs a unique function, executes in the correct order, and passes information seamlessly to the next, enabling multiple instructions to complete with minimal manual effort.
Key capabilities:
- Coordinate multiple agents across connected apps in a single instruction.
- Execute tasks sequentially with automatic handoff between agents.
- Edit or delete individual actions before starting the flow.
- Review and approve outputs (such as draft emails) before final execution.
Use when: You need to chain tasks across multiple apps or automate a multi-step workflow in one request.
Agentic Flow Example
Instruction: “Get my Jira issues. Let me know John Doe’s availability for a meeting tomorrow. Send the Jira issues to John Doe.”
Agents involved:
- Jira: Fetches the user’s Jira issues.
- Google Calendar: Checks John Doe’s availability for a specified time.
- Gmail: Sends the fetched Jira issues to John Doe via email.
Steps to execute the flow:
- Enter your instructions and press Enter. Each action maps to its respective agent.
- Hover over any instruction to display the Edit and Delete icons to modify or remove individual actions.
- Select Start to initiate the flow. Tasks execute sequentially.
- The first two actions run automatically.
- For the third action, the platform creates a draft email. Review and edit the draft, then select Send to complete the task.
