# AssemblyAI MCP connector The AssemblyAI connector lets Claude, ChatGPT, Cursor, or any MCP client upload call recordings, transcribe them, search the words spoken, and pull summaries and subtitles from AssemblyAI, with every request logged by Elaichi. Source: https://elaichi.ai/connectors/assemblyai/ ## Facts | | | | --- | --- | | Application | AssemblyAI | | Category | Conversational Intelligence | | AI tools | 14 | | Authentication | Connects with an API key | | Needs your own OAuth app | No | | MCP endpoint | https://api.elaichi.ai/mcp | | Works with | Claude, ChatGPT, Cursor, any MCP client, and the Elaichi Agent | | Tools advertised by name | Yes | ## What you can ask once AssemblyAI is connected - List last week's AssemblyAI transcripts and their durations. - Pull the paragraphs from yesterday's customer call transcript. - Search transcripts for mentions of pricing and renewal. ## Connect AssemblyAI in Elaichi This happens once for the organization, before any client is involved. 1. Open Connections, choose Add connection, and pick AssemblyAI. 2. Optionally set Share with, then press Connect. 3. Paste an AssemblyAI API key. One person generates a token in AssemblyAI and pastes it once. Everyone else works through Share with, and never sees it. Credentials are vaulted and nobody, including the AI, reads them back. The connection becomes a toolbox immediately, so you can curate which AssemblyAI tools are exposed, rename them, or freeze arguments before anyone points a client at it. ## Connect AssemblyAI to Claude Endpoint: https://api.elaichi.ai/mcp 1. Open Customize, then Connectors. 2. Press Add. 3. Name it, paste the MCP server URL, then Continue. 4. Sign in and approve. On Team and Enterprise, an Owner adds it once. Everyone else turns it on for themselves. ## Connect AssemblyAI to ChatGPT Endpoint: https://api.elaichi.ai/mcp 1. Open Plugins, then press the + button. 2. Name it and paste the endpoint into Server URL. 3. Leave Authentication on OAuth, then tick the risk acknowledgement. 4. Press Create, then sign in and approve. Works on the web today. The plugin directory lives at chatgpt.com/plugins. ## Connect AssemblyAI to Cursor Endpoint: https://api.elaichi.ai/mcp 1. Open `~/.cursor/mcp.json`. 2. Add the endpoint under `mcpServers`. 3. Reload Cursor, then sign in and approve. Set up per machine, so repeat it on each computer you work from. ## Connect AssemblyAI to any MCP client Endpoint: https://api.elaichi.ai/mcp 1. Add the endpoint as a remote MCP server. 2. Sign in and approve. The Elaichi Agent already has these tools, with nothing to set up. ## What the consent screen decides Only Read is granted by default, which is not enough to call a AssemblyAI tool. Over MCP there is no trusted place to confirm a write in the moment, so the consent screen is the standing approval rather than a formality. Grant Read and Run tools. Think hard before granting Delete, which reaches into connected apps and cannot be undone. ## What teams do with AssemblyAI through Elaichi ### Turn a discovery call into notes Sales. Upload the recording after a call, get the transcript back, and ask for the objections, next steps, and who promised what, ready to paste into the deal. ### Find every call that mentions a problem Support. Search across transcribed support calls for a product name or an error a customer described and pull the matching sentences with timestamps. ### Get subtitles for a webinar recording Marketing. Transcribe the recording and fetch the subtitle file in the format your video platform expects, then tidy the wording before it goes live. ### Share a call with card numbers removed Compliance. Pull the redacted version of a recording so a coaching session or a vendor review never hears the card number or address a customer read out. ### Summarize a batch of customer interviews Research. Transcribe a week of interviews, split each into paragraphs, and ask for the themes that came up more than once across the whole set. ### Clear out old transcripts on schedule Operations. List transcripts older than your retention period and delete them, so recordings of past hires and former customers are not kept longer than policy allows. ## Frequently asked questions ### How do I connect AssemblyAI to Claude? In Elaichi, choose AssemblyAI and paste the API key from your AssemblyAI dashboard, which is the whole sign-in step, since there is no OAuth application to register and no client ID or secret to generate. Then open Claude, go to Customize, then Connectors, then Add, and paste https://api.elaichi.ai/mcp. Claude asks you to sign in to Elaichi as yourself, and AssemblyAI is available in your next conversation. ### Does AssemblyAI work with ChatGPT and Cursor as well as Claude? Yes. Once AssemblyAI is connected in Elaichi, the same endpoint, https://api.elaichi.ai/mcp, works in Claude, ChatGPT, Cursor, any other MCP client, and the Elaichi Agent. You connect AssemblyAI once and every client you use picks it up. ### What can an AI agent actually do with my AssemblyAI data? It can upload a recording to AssemblyAI, start a transcript, and fetch it back as full text, sentences, or paragraphs with timestamps. It can search transcripts for particular words, fetch a subtitle file, pull the redacted version of a recording, and answer questions about what was said, such as listing the action items from a call. It can also list and delete transcripts in your AssemblyAI account. ### Does connecting AssemblyAI give the AI access to every transcript in my account? An AssemblyAI API key covers the account it belongs to, so the AI sees the transcripts that account can already see, and nothing beyond that. Elaichi can narrow that further, for example by allowing transcription but not deletion, or by limiting who in the organization can use the AssemblyAI connection at all. Elaichi never widens what the key itself allows. ### Can my team share one AssemblyAI connection? Yes. One person connects AssemblyAI with the API key and shares the connection with a team in Elaichi, and nobody else on the team ever handles the key. Each teammate still signs in to Elaichi as themselves, so the audit log records which person transcribed, searched, or deleted a recording. ### Can I stop an agent from deleting transcripts in AssemblyAI? Yes. Restrictions in Elaichi apply per action, so you can allow uploading and transcribing in AssemblyAI while blocking deletion entirely. A blocked action is never advertised to Claude, ChatGPT, Cursor, or any other client, so no prompt, however worded, can reach it. ### What happens to an AssemblyAI connection when someone leaves? When you offboard a person in Elaichi, their access to AssemblyAI through every client ends at once. If they had shared the AssemblyAI connection with a team, it keeps working for everyone else on that team. If you want the connection gone, disconnecting AssemblyAI once in Elaichi removes it from Claude, ChatGPT, Cursor, and every other client at the same time. ## All 14 AssemblyAI tools Every tool below is callable through https://api.elaichi.ai/mcp once AssemblyAI is connected, subject to the toolbox it is in and the restrictions on the caller. - **Create a AssemblyAI upload** (Create). Upload a media file to AssemblyAI's servers as raw binary data. Returns the uploaded file object upon success, which can be referenced in subsequent transcription requests. - **List all AssemblyAI transcripts** (List). List AssemblyAI transcripts sorted from newest to oldest. Returns: id, status, audio_url. Transcripts are available for the last 90 days of usage. - **Create a AssemblyAI transcript** (Create). Create an AssemblyAI transcript from a media file accessible via URL. Returns the transcript object including id and status; processing is asynchronous—poll via get until status is "completed". Required: audio_url. - **Get single AssemblyAI transcript by ID** (Get). Get an AssemblyAI transcript by id. Returns the transcript resource including id, status, text, and audio_url; the transcript is ready when status is "completed". Required: id. - **Delete a AssemblyAI transcript by ID** (Delete). Delete an AssemblyAI transcript by id, removing all associated data and marking it as deleted. Returns the deleted transcript object. Required: id. - **List all AssemblyAI sentences** (List). Get the sentences for an AssemblyAI transcript, semantically segmented for reader-friendly output. Returns an array of sentence objects. Required: transcript_id. - **List all AssemblyAI paragraphs** (List). Get the paragraphs of an AssemblyAI transcript, semantically segmented for reader-friendly output. Returns an array of paragraph objects. Required: transcript_id. - **List all AssemblyAI word searches** (List). Search an AssemblyAI transcript for keywords or phrases and return all matching occurrences. Returns: matches. Required: transcript_id, words. Each search term can be an individual word, number, or phrase of up to five words. - **List all AssemblyAI redacted audios** (List). Get the redacted audio in AssemblyAI for a specific transcript using transcript_id. Returns status indicating redacted audio readiness and redacted_audio_url containing the downloadable file link. Redacted audio is available for 24 hours only. - **Get single AssemblyAI subtitle by ID** (Get). Get subtitles for an AssemblyAI transcript in the specified format. Returns the subtitle file content for the given transcript. Required: id, subtitle_format. - **Create a AssemblyAI chat completion** (Create). Create a chat completion in AssemblyAI. Generates model responses based on provided messages or prompt. Returns choices array with message content, finish_reason, and usage details for token counts. - **Create a AssemblyAI speech understanding** (Create). Create a speech understanding task in AssemblyAI for a given transcript_id. Performs translation, speaker_identification, or custom_formatting on the transcript. Returns response objects with task status, mappings, formatted_text, and translated_texts. - **List all AssemblyAI streaming token** (List). Generate a temporary streaming_token in AssemblyAI. Requires expires_in_seconds parameter. Returns token and expires_in_seconds fields, which indicate the generated temporary authentication token and its redemption window. - **List all AssemblyAI voice agent token** (List). Generate a temporary Voice Agent token in AssemblyAI. Requires expires_in_seconds. Returns token and expires_in_seconds fields used to authenticate a single session.