> ## Documentation Index > Fetch the complete documentation index at: https://docs.retellai.com/llms.txt > Use this file to discover all available pages before exploring further. # Retell AI Conductor overview > Use Retell AI Conductor to build agents, improve analytics dashboards, investigate conversations, and open resources while keeping your work in context. Conductor is Retell’s AI assistant for building agents, investigating conversations, and improving analytics dashboards. Ask for what you need in plain language, attach relevant context, and review proposed changes before applying them. Use Conductor to build and test agents, inspect knowledge bases, investigate calls and chats, and analyze dashboards, QA results, or contacts. The [functionality guides](#where-to-go-next) cover each workflow. Patient Screening template agent in the current builder with the Conductor panel. ## How to open Conductor * **From an agent:** click **Conductor** in the agent builder toolbar. * **From Analytics and other dashboard pages:** click **Ask Conductor** to open the panel alongside the page. * **From Home:** start typing in the Conductor message box; no agent needs to be open. Drag the panel’s left edge to adjust its width. You can also drag the floating **Ask Conductor** button along the bottom of the page to move it out of the way. Conductor stays available alongside open dialogs and detail panels. Close the panel when you need more room; your conversation stays in history. Starter suggestions change with the page you have open. Click a suggestion to fill the message box, edit it if needed, then send it. ## Work across dashboard pages Conductor is available on every dashboard page and uses the current page as context. You can [attach other resources](#give-conductor-context) to compare agents or investigate conversations across your workspace. Conductor’s replies can include navigation cards with a **View** button. Click **View** to open the resource and continue the same conversation there. For example, ask "Show unsuccessful calls from last week" to get a **Call History** link with those filters applied. A request to show unsuccessful calls from last week, followed by Conductor’s Call History navigation card filtered to September 7–13, 2026, with a View button. Cards can open agents and specific versions, filtered call or chat history, analytics dashboards with filters and breakdowns, or AI QA cohorts. Check the destination’s filters to confirm the selection. Replies can also include **inline links** to individual calls, chats, and conversation-flow nodes. Click a call or chat link to open its details, or a node link to focus it in the agent builder, while keeping the conversation open. ## Give Conductor context Attach resources or files when your request concerns a specific item. They appear as **context chips** above the message box. For example, attach two appointment-booking agents and ask how they handle rescheduling differently. ### How to attach context Click **Add context** (+) in the Conductor message box. Choose a category, such as **Agents** or **Call History**, then click an item to attach it. To attach several items from a category, click **Multi-select**, check the items, and click **Save**. Repeat for other categories as needed. Check the context chips above the message box, type your request, and send it. Conductor Add context selector beside the message box, with Nodes selected, Greetings and Wrong User checked, and the Save button. Larger selections of the same resource type appear as a single chip with a count. Your sent messages keep their attachment chips, so you can see which context accompanied each request. ### What can you attach? The available categories depend on your permissions and the page you have open. | Category | What it provides | Availability | | ---------------------------- | --------------------------------------------------------------------- | ------------------------------------ | | Agents | Agents to reference or compare. | Any page | | Knowledge Base | [Source metadata and indexing status](/conductor/knowledge-base). | Any page | | Call History / Chat History | Conversations to investigate. | Any page | | Analytics | An [analytics dashboard](/conductor/analytics-dashboards) to discuss. | Any page | | AI QA | A [cohort’s evaluation results](/conductor/ai-qa). | Any page | | Contacts | A [contact and its conversation history](/conductor/contacts). | Any page | | Nodes | Conversation-flow nodes. | Conversation-flow agent | | Test Cases / Testing History | Saved test cases and test runs. | Agent page, except custom LLM agents | | Charts | Individual charts. | Analytics dashboard | With Conductor open in a conversation-flow agent, you can also click a node on the canvas to attach it. For a targeted edit, attach the "Collect email" node and ask, "Make this step confirm the spelling back to the caller." To [investigate a failed test](/conductor/test-and-improve), attach its test case or run and ask what needs to change. ### Select text anywhere in the UI Highlight text anywhere in the dashboard to give Conductor context about what you’re reading. For example, highlight **Voice & Language** in an agent’s settings and ask, "Explain how this setting affects my agent." The selection appears as a character-count chip above the message box. Check that the chip appears before sending. If it says **Selection too large**, select a smaller passage. Voice & Language text highlighted in the agent settings beside Conductor, where the resulting 16 characters chip is above a question about the selected setting. ### Select calls and chats from history On **Call History** or **Chat History**, click **Select**, check the conversations, and open Conductor. The selected rows appear as context chips. Ask, for example, "What do these conversations have in common?" Call History in selection mode with two checked web-call rows and a 2 selected count. See [Review calls and chats](/conductor/call-and-chat-history#find-and-review-conversations) for the full workflow. ### Attach files Click the **paperclip** in the message box to upload files, or **drag and drop files** directly into the message box. You can also paste an image from your clipboard. **Supported file types:** images (JPG, JPEG, PNG, GIF, WebP), PDF, and text-based files (TXT, CSV, HTML, HTM, MD, RTF, JSON). **Limits:** up to 10 files per message, each 50 MB or smaller (as of September 2026). Conductor identifies files it cannot attach because of their type, size, or the file-count limit. For example, attach your appointment policy and ask Conductor to use it when [refining your agent](/conductor/build-and-refine). ## Review proposed changes Conductor previews proposed agent and dashboard edits for your review. Compare the current and proposed values, accept or reject each change, then submit. See [agent change review](/conductor/build-and-refine#review-and-apply-changes) or [dashboard proposals](/conductor/analytics-dashboards#build-or-improve-a-dashboard) for the steps. If a proposal belongs to another page, click **Open this agent** or **Open this dashboard** to review it there. ## Follow a guided walkthrough Ask Conductor how to do or find something in the dashboard, such as "Show me where to change this agent’s voice." Conductor can start an interactive walkthrough that highlights the steps on the page. Follow the highlighted instructions to complete the task. Two-step dashboard walkthrough: first, open Global Settings; then, locate the current Cimo voice control. The guide displays step progress and Back and Got it controls without changing the agent’s settings. Depending on your question, Conductor may answer directly or ask for more detail before starting a walkthrough. If a walkthrough is unavailable or cannot start, it offers another way to help. ## Manage your conversations Your conversation history is shared across dashboard pages, so you can reopen a chat and continue where you left off. ### Start a new conversation Click **New chat** at the top of the panel when you switch tasks. The previous conversation stays in history. Conductor header with new chat, history, and close controls. ### Open your chat history Click **History** to reopen a conversation. On Home, use the searchable **Conductor History** panel. **Current Page** marks chats last used on the page you have open. Conductor restores your most recent chat when you reopen the panel. If its last message was sent elsewhere, **Last used on** links back to that page with the conversation open. ### Rename or delete a conversation Open the conversation title menu and choose **Rename chat** or **Delete chat**. Enter a new title and press **Enter**, or confirm deletion when prompted. You can also hover over a history entry to open its rename/delete menu. Deletion cannot be undone. ### Give feedback Click **Helpful** (👍) or **Not helpful** (👎) on a reply. After **Not helpful**, describe what you expected and click **Submit**, or choose **Skip** to leave only the rating. Click a selected thumb again to clear it, or choose the other thumb to change it. Open the conversation title menu and click **Copy chat ID** to identify it for support. When you open a resource through a reply’s navigation link, the destination URL includes the conversation. Copy that URL to revisit the same resource and chat, or share it with a teammate who has workspace access. Conductor automatically condenses earlier context in long conversations. You may briefly see **Condensed conversation context** in its activity. Continue in the same chat after condensation, or start a **New chat** for a different task. ## Pricing Conductor includes a daily free message allowance. Workspace admins can enable pay-as-you-go and set a monthly spending limit. See [Pricing](/conductor/usage-and-billing) for allowances, reset times, and billing details. ## Where to go next Create and refine agents, configure functions, and review changes. Create test cases, run simulations, and investigate failures. Inspect sources, indexing status, and agent associations. Find conversations and investigate transcripts, tool calls, and outcomes. Build dashboards, discuss charts, and review analytics changes. Inspect cohort trends and check individual call evaluations. Find contacts and review their conversation history across agents.