# Observe.AI MCP > Observe.AI MCP connects your AI agent directly to your contact center performance data. Get instant visibility into call transcripts, quality assurance scores, and coaching notes without leaving your workspace. Analyze every interaction—from greetings to objections—and track agent improvements using natural language prompts. ## Overview - **Category:** customer-support - **Price:** Free - **Tags:** conversation-intelligence, contact-center, quality-assurance, speech-analytics, call-transcription, performance-coaching ## Description Connect this MCP to gain deep insight into how your customer service teams perform. You don't have to open the Observe.AI portal or manually search through spreadsheets anymore. Your AI client pulls performance data directly, allowing you to ask complex questions like, 'What was the average QA score for agents who handled billing issues last week?' The system collects everything—from full conversation transcripts to automated summaries and coaching feedback logs—and presents it in plain language. By using Vinkius, your agent gets access to this entire catalog of tools, letting you query calls, chats, and emails all from one place. This means QA Analysts can quickly check evaluation scores; Managers can monitor high-level trends during daily standups; and Coaches can verify improvement history instantly. ## Tools ### get_evaluation_details Retrieves specific quality assurance evaluation details for an interaction. ### get_interaction_details Gets general metadata and information about a specific customer interaction. ### get_interaction_transcript Pulls the full text transcript of a recorded call or chat conversation. ### list_coaching_sessions Lists all documented coaching sessions for a specific agent. ### list_qa_evaluations Retrieves a list of all available quality assurance evaluations. ### list_evaluation_forms Lists the specific forms used for QA evaluation. ### list_interactions Retrieves a list of recent contact center interactions, including calls and chats. ### list_interaction_moments Lists key business moments (like 'Greeting' or 'Objection') identified by the AI within interactions. ### list_interaction_summaries Provides a list of automated, high-level summaries for recent interactions. ### list_workspace_users Retrieves a directory listing of agents and administrative users in the Observe.AI workspace. ## Prompt Examples **Prompt:** ``` List all recent call interactions from today. ``` **Response:** ``` Retrieving interactions... I found 12 calls from today. Notable ones include 'Interaction #98765' (Duration: 5:30, Sentiment: Positive) and 'Interaction #98766' (Sentiment: Frustrated). Would you like the transcript for the frustrated call? ``` **Prompt:** ``` What is the QA score for interaction ID 'int_12345'? ``` **Response:** ``` Fetching evaluation details... The interaction int_12345 received a total QA score of 95%. It scored perfectly on 'Compliance' and 'Empathy', with a minor deduction in 'Resolution Speed'. Shall I list the full evaluator feedback? ``` **Prompt:** ``` Show me the AI summaries for our latest interactions. ``` **Response:** ``` Retrieving AI summaries... I found 5 recent summaries. The overall theme focuses on 'Renewals' and 'Billing Inquiries'. One summary identifies a recurring issue with 'Portal Login'. Would you like the detailed summary for that specific case? ``` ## Capabilities ### Review all service interactions List every call, chat, or email processed by the platform, along with metadata. ### Retrieve full conversation text Pull the complete text transcript for any specific interaction so you can review details instantly. ### Assess agent quality scores Access formal quality assurance evaluation forms, individual scores, and performance metrics. ### Identify key conversation moments List specific business moments identified by the AI, such as greetings or customer objections, across multiple interactions. ### View summarized call themes Read automated summaries that distill the main topics discussed in recent conversations. ### Track coaching history List and review records of agent coaching sessions and feedback given by supervisors. ## Use Cases ### Investigating a bad customer experience. A manager hears about a repeat complaint. They ask their agent to pull the full transcript using get_interaction_transcript for all interactions involving that customer in the last week. This immediately highlights patterns and shows exactly where the service broke down. ### Preparing for team reviews. A coach needs proof of improvement. They use list_coaching_sessions to pull a history of past feedback, then ask the agent to compare that progress against recent QA scores using get_evaluation_details. ### Quickly understanding market shifts. The team needs to know if billing issues are spiking. They prompt the agent for list_interaction_summaries, which quickly reveal 'Billing Inquiries' as the dominant theme across all interactions this month. ### Auditing compliance failures. A QA analyst suspects a team is missing mandatory disclosures. They use list_interaction_moments to filter for instances where the required 'Compliance Statement' was not recorded, providing actionable data points. ## Benefits - Stop hunting for transcripts. Use get_interaction_transcript to pull the full text of any chat or call in seconds, letting you review conversations without leaving your workflow. - Track performance trends instantly. Rather than digging through dozens of reports, list_qa_evaluations gives you a clean summary of quality scores across teams and time periods. - Automate daily reporting. Use list_interaction_summaries to get immediate high-level overviews of what customers are talking about—great for quick manager updates. - Spot training gaps fast. By calling list_coaching_sessions, supervisors can see exactly when an agent was coached and on what topics, proving progress or identifying recurring weakness. - Go deeper than scores. Calling list_interaction_moments lets you pinpoint *why* a call failed—did they miss the 'Closing' moment? Did they fail to acknowledge the customer's 'Objection'? - Manage personnel data easily. Use list_workspace_users to quickly verify who is on your team and who needs access, all without logging into the main admin portal. ## How It Works The bottom line is you get instant answers about your contact center performance without logging into multiple dashboards or systems. 1. Subscribe to this MCP on Vinkius and input your Observe.AI API Key (Bearer Token). 2. Connect your preferred AI client, like Cursor or Claude, to the catalog. 3. Ask your agent a question—for example, 'What were the top three objections raised by customers last month?'—and it pulls the data. ## Frequently Asked Questions **How does Observe.AI MCP handle transcripts?** You can retrieve the full text transcript for any call or chat interaction by calling get_interaction_transcript. This gives you the complete conversation history immediately in your agent's response. **Can I check historical QA scores using Observe.AI MCP?** Yes, you can list all available quality assurance evaluations using list_qa_evaluations to see a record of past scoring efforts and trends. **What is the best way to analyze agent performance with Observe.AI MCP?** Start by listing interactions using list_interactions, then ask for get_interaction_details on any specific ID. This gives you core metadata necessary to understand context before diving into scores. **How do I find out what customers are complaining about?** Ask the agent to use list_interaction_summaries or list_interaction_moments. These tools automatically identify recurring themes and key moments like 'Objection' across many calls. **Does Observe.AI MCP help with coaching records?** Yes, you can use the list_coaching_sessions tool to pull a history of agent training sessions and track when specific feedback was given.