# Retell AI MCP > Retell AI MCP lets you build human-like voice agents that take over phone calls or web conversations. Your agent can answer questions, complete tasks, and manage complex workflows using natural spoken language. You gain full control of call lifecycle management—from initiating calls to reviewing detailed transcripts and monitoring performance metrics—all through your preferred AI client. ## Overview - **Category:** industry-titans - **Price:** Free - **Tags:** voice-ai, conversational-ai, telephony, automated-calls, voice-agents, llm-integration ## Description Need an automated voice layer for your business? This MCP connects the power of Retell AI directly to any agent, letting you build sophisticated agents that handle phone or web interactions using natural speech. Instead of manually setting up complex call flows or sifting through separate CRM logs, you simply instruct your AI client what you need done. You can programmatically create new personas and list existing ones, making sure every voice agent is set up correctly. When a call happens, whether it's an outbound sales pitch or a customer support chat, the MCP initiates and tracks the entire lifecycle. After the conversation ends, you instantly access detailed transcripts and all relevant metadata right where you're working. If your company operates on phone calls, this Vinkius catalog tool gives you total visibility into your voice operations without ever leaving your chat window. ## Tools ### get_phone_number Retrieves specific details for a given phone number. ### create_voice_agent Builds and deploys a new AI voice agent with defined capabilities. ### start_phone_call Initiates an outbound call to a specific phone number. ### start_web_call Starts a real-time, browser-based conversation session for the agent. ### get_agent_config Fetches specific configuration details for an existing voice agent. ### get_call_details Retrieves a full transcript and metadata report for any completed call. ### get_llm_details Gets the current settings and knowledge base details for an agent's response engine. ### list_voice_agents Lists all active AI voice agents currently managed by your account. ### list_recent_calls Provides a summary and list of call logs from the most recent interactions. ### list_retell_llms Lists all internal response engines available for agent use. ### list_retell_numbers Shows a comprehensive list of all registered phone numbers in your system. ## Prompt Examples **Prompt:** ``` List all my voice agents in Retell AI. ``` **Response:** ``` I've retrieved your agents. You have 3 active personas: 'Customer Support Cimo', 'Sales Assistant', and 'Outbound Lead'. Which one would you like to update? ``` **Prompt:** ``` Show me all AI voice agents and their call statistics from the last 7 days. ``` **Response:** ``` 4 active voice agents. "Sales Qualifier" handled 234 calls (avg duration: 3.2 min, 67% qualified leads, 89% caller satisfaction). "Support Assistant" took 567 calls (avg 4.1 min, 78% resolved without transfer, 92% satisfaction). "Appointment Scheduler" booked 145 appointments from 189 calls (76.7% booking rate). "Survey Collector" completed 89 surveys from 120 calls (74.2% completion). Total calls: 1,010. Total minutes: 3,434. Average latency: 0.8 seconds. ``` **Prompt:** ``` Create a new outbound phone call using the Sales Qualifier agent to contact a prospect. ``` **Response:** ``` Outbound call initiated. Agent: Sales Qualifier (ID: agent_4521). Phone number: +1-555-0234 (from your verified number pool). Calling: +1-555-0198 (prospect: David Chen, ScaleUp Inc). Call script: Lead qualification with product demo scheduling. Call status: Ringing... Connected at 0:03. Agent is conducting the qualification flow. Live transcript available in dashboard. Expected duration: 3-5 minutes. Recording: enabled. Post-call webhook configured for CRM sync. ``` ## Capabilities ### Manage Agent Personas List all active agents and create new ones programmatically, giving them specific roles or knowledge bases. ### Execute Calls Start both outbound phone calls and web-based browser calls directly from your chat interface. ### Monitor Call History Review detailed call logs, including transcripts and metadata, for past interactions. ### Control Intelligence Settings Check the current configuration of the underlying LLMs that power your agents to ensure they use the right knowledge base or logic. ### Track Telephony Assets View a clear list of all available phone numbers used by your organization for calling. ## Use Cases ### Quality Assurance Review A QA analyst wants to check how the 'Support Assistant' handled a complex billing query last week. Instead of searching through call center software, they just prompt their agent with 'list_recent_calls' and then use 'get_call_details' on the transcript to analyze exactly what was said. ### Scaling Sales Outreach A sales director needs to launch a new qualification campaign. They first check available lines using 'list_retell_numbers', then use 'create_voice_agent' for the new persona, and finally prompt the agent to 'start_phone_call' to begin outreach. ### Debugging Agent Behavior A developer notices an agent is giving outdated answers. They check the current logic using 'get_llm_details', identify the issue, and then use a natural language command to adjust the agent's behavior immediately. ### Onboarding New Teams A new team needs an automated web chat assistant. The PM uses the MCP to 'start_web_call', tests the setup, and then documents the successful flow in their internal wiki, knowing they can easily repeat the process. ## Benefits - You instantly get call transcripts and logs without leaving the chat interface. Reviewing past interactions is fast, using tools like 'list_recent_calls' instead of jumping between separate dashboards. - Stop manually configuring agents. Use 'create_voice_agent' to build new personas with specific roles; you can then use 'get_agent_config' to tweak parameters instantly when needed. - Gain total visibility into your calling infrastructure by using 'list_retell_numbers', ensuring your team knows exactly which numbers are available for outreach campaigns. - Monitor the intelligence layer. Use 'get_llm_details' and 'list_retell_llms' to confirm that every agent is operating with the correct, up-to-date knowledge base. - Run live operations without friction. Initiate both phone calls ('start_phone_call') and web calls ('start_web_call') directly through your AI client for immediate testing or execution. ## How It Works The bottom line is you get centralized control over every aspect of your conversational voice operations. 1. Subscribe to this MCP and enter the Retell AI API key from your dashboard settings. 2. Use natural language prompts within your AI client to list or create voice agents, or initiate a new call. 3. The agent executes the action, providing real-time status updates, transcripts, or structured data about the outcome. ## Frequently Asked Questions **How do I start using the Retell AI MCP with phone calls?** You initiate a call by prompting the agent to 'start_phone_call'. The system handles the connection, and you receive real-time updates on the status and transcript right in your chat interface. **Can I list all my voice agents using Retell AI MCP?** Yes, simply ask to 'list_voice_agents'. This command retrieves a full roster of every active persona you've set up for your operations team. **What information does the get_call_details tool provide?** The 'get_call_details' tool provides everything: a complete transcript, call duration, and all relevant metadata about that specific interaction. **Does Retell AI MCP help me manage phone numbers?** Yes. You can use 'list_retell_numbers' to get an inventory of every registered number, giving you a clear overview of your entire telephony setup. **How do I ensure my agent has the correct knowledge base?** You should check and update this using 'get_llm_details' or 'list_retell_llms'. This confirms that the underlying response engine is running with the most accurate information.