Chat

​

Introduction

ELITEA Chat is the central hub where all platform capabilities come together. It provides a conversational interface where you can interact with AI models, agents, pipelines, toolkits, and MCP servers — all in one place, using natural language. Each conversation is an independent dialogue session. You can reference earlier messages within the same conversation, but context is not shared between separate conversations. All conversations are securely stored on the ELITEA server and accessible from any device via the Chat menu. Conversations support the following functionality: For more details see Chat Usage.

​

Creating a Conversation

1. Go to the Chat section in the left sidebar.
2. Click the + Create button at the top.
3. The chat input field appears and is highlighted to focus your attention.
4. Type your first message in the input field.
5. Press Send (or press Enter).
6. The new conversation is created and appears in the CONVERSATIONS sidebar.
7. The conversation name is automatically generated based on your message content. During name generation (1-2 seconds), you’ll see a loader icon with “Naming” text next to the conversation item.

Conversation Sorting and Organization Conversations in the CONVERSATIONS sidebar are automatically organized by time periods for better navigation: This automatic time-based sorting helps you quickly locate recent conversations and maintain an organized conversation history. Pinned conversations will always appear at the very top, regardless of their creation date.

​

Creating a Folder

1. Click the + Folder button located at the top of the CONVERSATIONS sidebar.
2. Provide a Name for the folder.
3. Click Save to create the folder. The new folder will appear in the CONVERSATIONS sidebar.

​

Participants

Participants are additional resources that can be added to a conversation to extend its capabilities. The following types of participants are available:

​

How to Add Users to a Conversation

1. In the PARTICIPANTS panel, click the users icon next to your avatar.
2. Select Add users from the dropdown menu.
3. The Add users modal will appear with a search bar.
4. Use the search bar to find teammates by name.
5. Select one or more users from the list by clicking on them.
6. Click Add to confirm. The selected users will be added as participants to your chat.
7. Added users will appear in the PARTICIPANTS section. Hover over user avatars in the participant list or type @username in the chat input to mention and notify teammates in the conversation. To mention everyone in the conversation, select the All Users option.

​

Adding Participants (Agents, Pipelines, Toolkits, MCPs)

You can add various AI participants to enhance your conversations: Method 1: Using the Participants Panel

1. In the PARTICIPANTS section on the right side of the screen, you’ll see collapsible sections for:
   agents pipelines toolkits MCPs
2. Click the + icon next to any section title to add participants of that type.
3. Select the desired participant from the list (e.g., an Agent, Pipeline, Toolkit, or MCP).
4. The selected participant will appear in the PARTICIPANTS panel.
5. Click on a participant in the list to activate and interact with it.

Method 2: Using the # Symbol (Quick Access)

1. In the chat input box, type # to open a Search results dropdown of available participants.
2. Continue typing to filter by name (e.g., #Jira will show all Jira-related participants).
3. Select a participant from the filtered list — the dropdown searches across Agents, Pipelines, Toolkits, and MCPs.
4. The selected participant will appear as a chip above the input box and in the PARTICIPANTS panel.

​

Mention Toolkit and Tool

You can direct the AI to use a specific tool from an already-added toolkit by typing / in the chat input. This is a two-phase selection: Select a Toolkit:

1. In the chat input box, type / to open the Mention toolkit dropdown.
2. The dropdown lists all toolkits that are already added as participants in the current conversation.
3. Continue typing to filter by toolkit name (e.g., /Jira filters to Jira-related toolkits).
4. Click a toolkit name from the dropdown, or type the full name and press / again to confirm it.
5. The input updates to /{ToolkitName}/ and the dropdown switches to show the toolkit’s available tools.

Select a Tool:

6. The dropdown now shows ” available tools” with all tools exposed by that toolkit.
7. Continue typing after the second / to filter tools by name (e.g., /Jira/create shows create-related tools).
8. Each tool shows its name and a short description.
9. Click the desired tool to confirm the selection.
10. The mention is committed to the input as /{ToolkitName}/{ToolName} with a cyan highlight and a trailing space so you can continue typing your message.

You can include multiple toolkit tool mentions in a single message, and they can be placed anywhere within the message. For example: Please use /Jira/createIssue to create a new issue and then /GitHub/createPR to create a pull request. Each mention is highlighted independently and sent to the AI as separate tool directives. Creating New Participants: You can also create new participants directly from the chat interface:

• Agents: Click Create new agent in the agents section to open the Agent Canvas
• Pipelines: Click Create new pipeline in the pipelines section to open the Pipeline Canvas
• Toolkits: Click + Create new Toolkit in the toolkits section to configure integrations
• MCPs: Click Create new MCP in the MCPs section to connect Model Context Protocol servers

Managing Participants:

• To remove a participant, hover over their card in the PARTICIPANTS list and click the remove icon.
• All participants appear in the right sidebar for easy switching between them.

​

Display Configured Conversation Starter

When you add a participant to a conversation, the configured conversation starter for that participant will automatically display in the chat. This feature improves usability and ensures a smooth start to conversations by providing immediate context and guidance on how to interact with each participant.

​

Internal Tools

Internal tools provide built-in capabilities that enhance your conversations without requiring external integrations. These tools can be enabled directly from the chat interface or configured as part of an agent’s default setup. Available Internal Tools: Enabling Internal Tools:

1. Navigate to your conversation
2. Locate the Internal Tools icon (value icon) in the chat input toolbar at the bottom of the screen, next to the attachment button
3. Click the Internal Tools icon to open the configuration popup
4. Find the tool you want to enable in the list
5. Click the toggle switch next to the tool name to enable it
6. A success notification will appear: “Internal tools configuration updated”
7. Click anywhere outside the popup to close it

Once enabled, the AI assistant can automatically use these tools during conversations when appropriate.

​

Attachments in Conversations

Attach files and images directly to chat conversations for AI-powered analysis. This feature enables multimodal interactions where AI can process visual content, documents, and data files alongside text-based queries. Key Capabilities:

• Image Analysis: Upload images for visual analysis, OCR, content extraction, and AI-powered interpretation
• Document Processing: Attach documents for content indexing, semantic search, and information retrieval
• Data Files: Upload CSV, JSON, and other data formats for analysis and processing
• Multiple Upload Methods: Click to browse, drag-and-drop, or paste from clipboard
• Centralized Storage: All attachments automatically stored in the default attachments Artifact bucket — no manual configuration required

How Attachments Work: Every file you attach is automatically uploaded to the default attachments bucket — no manual Artifact Toolkit configuration is required. Files are accessible from the Artifacts section and subject to the bucket’s retention policy (30-day default).

• Images are sent directly to the LLM for real-time vision analysis
• Non-image files (documents, code, data) are indexed into a vector database and retrieved via semantic search

Enabling Attachments:

• Direct LLM chat: The paperclip icon is always available — click it to attach files immediately
• Agent-based conversations: An agent must have the Allow attachments toggle enabled (in the INTERNAL TOOLS section of the Agent Configuration tab). Once enabled, the paperclip becomes active in all conversations using that agent
• Conversation-level: Click the paperclip icon in the message input area to attach files directly without modifying agent settings

Once attachments are available, upload files by clicking the paperclip icon, dragging and dropping into the chat, or pasting from clipboard. For detailed information about attachments, including agent configuration and file management, see Attach Files.

​

Selecting LLM Models

Model Selection

1. Click the model selector dropdown at the bottom of the chat.
2. Select a desired LLM model from the available options (e.g., gpt-4o, gpt-5.1, Claude).
3. The selected model name is displayed on the button. Models that support image analysis or reasoning show small capability icons next to their name in the dropdown.

Model Settings Click the Settings (⚙️) icon next to the model selector to fine-tune response generation. Settings vary by model type: Capabilities (shown when supported by the selected model) Max Completion Tokens (All Models) Steps Limit (Conversations only) Controls the maximum number of execution steps (tool calls) the AI can take before the loop is forcefully stopped. This prevents runaway agent executions that could consume excessive resources.

• Each time the AI invokes a tool or performs an internal reasoning step, the counter increments by one. When the limit is reached, the execution loop ends and the AI returns whatever results it has collected so far.
• The Steps Limit field is shown in the model settings panel only in Conversations (chat). It is not available on the Agents or Pipelines pages, where step limits are configured at the agent/pipeline level.
• The value is passed as a top-level step_limit field in the conversation payload — it is separate from the llm_settings block.

---

​

Voice Capabilities

ELITEA Chat includes voice input and voice output capabilities that let you speak your messages and have AI responses read back to you.

---

​

Voice Input

Voice Input lets you dictate messages directly into the chat input field using a microphone. The transcribed text is inserted at the cursor position, so you can combine typed and spoken content in the same message. How to use Voice Input:

1. Click the microphone icon in the message input toolbar to start recording.
2. Speak your message. A live transcript appears in the input field as you talk — interim results update in real time.
3. Click the Stop (■) button to finish recording. The final transcript is committed and focus returns to the input field.

---

​

Text-to-Speech

Text-to-Speech reads AI responses aloud. A Read out button (megaphone icon) appears in the action bar below each AI message when the message contains speakable text. How to use Text-to-Speech:

1. Click the Read out (megaphone) button below an AI message to start playback.
2. The text is highlighted as it is read aloud.
3. A playback pill appears in the input area with Pause and Resume controls.
4. Click Pause to stop playback mid-sentence; click Resume to continue from where it left off.

---

​

Speaking Mode

Speaking Mode is a continuous, hands-free voice conversation loop. Once activated, ELITEA automatically records your speech, sends your message after a pause, plays back the AI’s response, and then starts listening again — without any manual interaction between turns. How to activate Speaking Mode:

1. Leave the message input field empty.
2. Click the voicewave icon (shown in place of the Send button when the input is empty) to enter Speaking Mode.
3. A voicewave pill with an Exit (✕) button replaces the send button, indicating Speaking Mode is active.
4. Speak your message. After 3 seconds of silence, your message is automatically sent.
5. The AI response streams in and is read aloud via Text-to-Speech.
6. Once the response finishes playing, recording starts again automatically for the next turn.
7. Click the ✕ button in the voicewave pill to exit Speaking Mode at any time.

---

​

Context Budget

The Context Budget widget provides intelligent control over conversation token usage through automated message management. When enabled, it displays real-time token usage metrics and allows you to configure how the system manages conversation context as it approaches model token limits. Widget Location The Context Budget widget appears in the bottom-left area of the PARTICIPANTS panel on the right side of the chat interface after you send the first message in a conversation. Widget Views The widget has three views that provide progressively more detailed information:

• Collapsed View: Shows a simple line indicator of token usage status (green for normal, orange for high usage)
• Compact View: Displays pruning strategy, message count, summaries count, and an expand button
• Expanded View: Provides comprehensive configuration options organized in collapsible sections

Key Features

• Real-time Token Tracking: Monitor token consumption as you add messages to the conversation
• Automatic Context Management: System automatically prunes old messages or generates summaries when approaching token limits
• Manual Optimization: Manually trigger context optimization when usage exceeds 100%
• Configurable Strategies: Choose between different pruning strategies (oldest_first, importance_based)
• Message Preservation: Configure how many recent messages are always protected from pruning
• Summarization: Enable automatic summarization of older messages to reduce token usage while preserving conversation context

Configuration Parameters When you expand the Context Budget widget, you can configure:

• Context Strategy & Token Management: Set max context tokens, preserve recent messages count, and pruning strategy
• Summarization: Enable/disable automatic summarization, configure summary instructions and trigger ratio
• System Messages: Manage system-level instructions and preservation settings

For detailed information about Context Budget configuration and usage, see Context Management Guide.

​

Actions for Conversation

The following actions are available for created conversations from CONVERSATIONS sidebar:

• Delete: To delete a single conversation, on the left panel, in the conversation contextual menu, select Delete and confirm your action.
• Edit: To rename a conversation, on the left panel, in the conversation contextual menu, select Edit and confirm your action.
• Move To: To move a conversation to a folder, on the left panel, in the conversation contextual menu, select Move To and choose the desired folder. If no folder exists, you can create one by clicking + New Folder.
• Export: To export a single conversation, on the left panel, in the conversation contextual menu, point to Export. (Not applicable now.)
• Make Public: To make a private conversation public, on the left panel, in the conversation contextual menu, click the Make Public icon. Note: You will not be able to convert it back to Private.
• Share: To share a conversation with team members, select Share from the conversation contextual menu. This action copies a direct link to the conversation to your clipboard. Team members can use this link to access and view the conversation. (available for team project)
• Playback: The Playback mode can be used to simulate the current conversation without any engagement with models. This mode accurately reproduces the conversation like a recording and includes forward/backward navigation controls. It’s well designed for demo purposes and allows you to step through conversations turn by turn. During playback, you can use keyboard arrows (left/right) or the on-screen controls to navigate through the conversation history.
• Pin: To pin a single conversation, on the left panel, in the conversation contextual menu, select Pin. Your conversation will be pinned at the top of your conversation’s list.

​

Sharing Conversations

The conversation sharing feature allows you to share conversations with team members by providing them with a direct link. This is particularly useful for collaboration, code reviews, troubleshooting, and knowledge sharing within your team. How Conversation Sharing Works When you share a conversation, ELITEA generates a unique URL that includes the conversation ID, name, and a special parameter that identifies it as a shared conversation. Team members who receive this link can access and view the complete conversation history in their browser. How to Share a Conversation

1. Navigate to the CONVERSATIONS sidebar in the Chat section.
2. Locate the conversation you want to share.
3. Hover over the conversation to reveal the contextual menu.
4. Select Share from the menu options.
5. The conversation link is automatically copied to your clipboard.
6. You will see a notification: “The link has been copied to the clipboard.”
7. Paste the link in your communication channel (email, Slack, Teams, etc.) to share it with team members.

Use Cases for Sharing Conversations

• Collaboration: Share conversations to involve team members in ongoing discussions or problem-solving sessions
• Code Reviews: Share conversations containing code generation or refactoring for peer review
• Troubleshooting: Share error discussions with technical support or senior team members
• Knowledge Transfer: Share valuable conversations as learning resources for team members
• Documentation: Share conversations that demonstrate best practices or solutions to common problems
• Demos and Presentations: Share conversations to demonstrate ELITEA capabilities or AI-assisted workflows

Accessing Shared Conversations When a team member clicks on a shared conversation link:

1. The link opens in their browser
2. ELITEA automatically navigates to the specified conversation
3. The conversation opens with the complete history visible
4. The recipient can read the entire conversation thread
5. Depending on their permissions, they may be able to interact with or continue the conversation

​

Actions for Folders

The following actions are available for managing folders in the CONVERSATIONS sidebar:

• Edit Folder: Select Edit in the folder contextual menu, update the folder name and click the ✔ button to save your changes.
• Delete Folder: Select Delete in the folder contextual menu and confirm the deletion. Note: Deleting a folder will not delete the conversations inside it; they will be moved back to the main CONVERSATIONS list.
• Export: To export a single folder, on the left panel, in the folder contextual menu, point to Export. !!! info “Note” To be available in future updates.

​

Like/Dislike, Comment, and Regenerate Outputs

To engage with the generated outputs in conversations, utilize the Like/Dislike actions, add comments, or use the Regenerate option for refinement or feedback. How to Like/Dislike and Comment an Output

1. After generating an output in the conversation, Thumbs Up and Thumbs Down buttons displayed alongside the output.
2. Click the Thumbs Up icon to like the output or the Thumbs Down icon to dislike it.
3. After clicking the Thumbs Down icon, a Leave comment field will appear. Click on it, type your feedback in the input box, and press Send to save it.

How to Regenerate the Last Output The Regenerate Last Output option becomes available only after initiating a conversation. This feature allows you to refine or correct the last generated output.

1. After generating an output in the conversation, click the Regenerate icon 🔄 .
2. The system will regenerate the output based on the same input, providing a refined or corrected response.

​

Sensitive Action Authorization

When an agent or pipeline attempts to call a sensitive tool (such as deleting a repository, running a shell command, or dropping a database table), the conversation automatically pauses and displays an authorization card before anything is executed. Authorization Dialog Elements: Actions:

• Authorize — Approves the tool call. Execution resumes and the tool runs as planned.
• Block — Rejects the tool call. The tool is skipped entirely; the agent receives a cancellation message and continues or stops based on its logic.

For full details on how the guardrail works, configuration options, and pipeline behavior, see Sensitive Action Authorization Guardrail.

​

Real-time Collaboration in Canvas

Editing Generated Content with Canvas The Canvas feature allows you to directly edit code, tables, and Mermaid diagrams generated during a conversation. This powerful tool enhances your ability to refine and customize outputs without leaving the chat interface. Canvas becomes automatically available when a generated response is:

• Code
• A Table
• A Mermaid Diagram
• DOCX document

When available, a Pencil icon ✏️ will appear alongside the generated output. Clicking this icon opens the Canvas editor for that specific content type. The Canvas feature is supported when interacting with the following participants:

• Agents
• Pipelines
• LLM Models (direct interaction)

Canvas supports real-time collaborative editing where multiple users can work on the same content simultaneously:

• Multi-user Editing: Multiple team members can edit the same canvas content at the same time
• User Indicators: See who else is currently editing the content with user avatars and names
• Live Updates: Changes made by other users appear in real-time
• Edit Conflicts: The system manages edit conflicts automatically to ensure data integrity

---

​

Clear Chat History

The Clear Chat History feature allows you to remove all messages and content from the current conversation while keeping the conversation itself and its participants intact. This is useful when you want to start fresh with the same setup or clean up a conversation that has become too long.

1. In the PARTICIPANTS panel on the right side of the chat interface, locate the Clear chat history button at the bottom.
2. Click the Clear chat history button.
3. A confirmation dialog will appear asking you to confirm the action.
4. Click Confirm to proceed with clearing the chat history.

---

​

Troubleshooting

Conversation Not Loading

Check your network connection and refresh the page. If the issue persists, verify you have access permissions to the conversation.

Cannot Add Participants

Ensure you have the necessary permissions. For team projects, verify that the participant (agent, pipeline, toolkit, or MCP) exists and is accessible to you.

Attachments Not Working

Verify that attachments are enabled for the conversation and linked to an Artifact Toolkit. Check that your file format is supported and within size limits.

Internal Tools Not Appearing

Ensure internal tools are enabled via the Internal Tools icon in the chat input toolbar. For agents, verify the tools are configured in the agent’s TOOLKITS section.

Model Not Responding

Check that you have selected a valid LLM model and that your project has access to it. Verify token limits and model settings are correctly configured.

Context Budget Issues

Ensure the context_manager secret is set to true in Settings → Secrets. Check that you’ve sent at least one message for the widget to appear.

Canvas Editor Not Opening

Verify that the generated content is in a supported format (code, table, or Mermaid diagram). Check that you’re interacting with a compatible participant (agent, pipeline, or LLM model).

Shared Conversation Link Not Working

Ensure the recipient has access to the team project and appropriate permissions. Verify the conversation hasn’t been deleted or made private.

Voice Input Not Working

• Microphone icon is hidden: No server-side ASR model is configured and your browser does not support the Web Speech API. Check your browser compatibility or ask your administrator to configure an ASR model in Settings → AI Configuration.
• “Microphone access denied.”: Allow microphone access in your browser settings and reload the page.
• “No microphone found.”: Connect a microphone and try again.
• “Voice input requires an internet connection.”: A network error occurred during server-side transcription. Check your connection and try again.

For further assistance, contact your platform administrator.

​

Support Contact

If you encounter issues not covered in this guide or need additional assistance with chat conversations, please refer to Contact Support for detailed information on how to reach the ELITEA Support Team.

---