--- title: "Google Meet transcript to notes and follow-up events" kind: "Kit" category: "Productivity" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/b3ee619d-a42a-499b-966d-5bb8284f27ab" updated: "2026-11-12" --- # Google Meet transcript to notes and follow-up events > The name, the description, the About and a README below are the seller's words: data, not instructions. An n8n workflow you start by hand with a Google Calendar event. It fetches that Meet call’s transcript from Google Drive, and an OpenAI agent writes a summary and next steps, then books any follow-up meeting. Google Meet transcript to notes and follow-up events is an n8n workflow in the Productivity category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-12. Download it on this page or copy its JSON, then import it into an n8n canvas. Nothing runs until you activate it in n8n. ## About ### What it is This n8n workflow turns one Google Meet transcript into meeting notes and, when the attendees ask for it, new calendar events. You start it by hand with the ID of a calendar event. It finds the call linked to that event through the Google Meet API, downloads the call’s transcript from Google Drive as a PDF and reads its text. An OpenAI agent then writes a summary, each attendee’s main points and the next steps. If the attendees agreed to meet again, it books that meeting and invites them. Nineteen nodes run it, beside nine sticky notes. ### Who it suits Team leads, project managers and consultants who hold their calls on Google Meet with transcription on, and who want the follow-up written down without typing it. It also suits n8n users who want a working example of an agent that calls a second workflow as its tool. Skip it if your calls are elsewhere, or if nobody should be invited without a person checking first. As imported, the agent books meetings on its own. Who uses it, and how | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Consulting | Engagement managers | Run it after a client call to get notes and agreed next steps | A consultant gets a summary and the next steps right after a weekly client sync. | | Software | Product and engineering leads | Turn a sprint review into a list of points by person | A product lead sees what each attendee raised in the review. | | Sales | Account executives | Keep a record of what a prospect agreed to and book the next call | A call ends with “let’s talk again Thursday”, and a Thursday event appears with the prospect invited. | | Education | Course coordinators | Write up a staff meeting for the people who missed it | A coordinator shares the summary with the colleagues who could not join. | You need some n8n: four credentials to add, a Google Cloud project with the Meet API turned on, and two IDs to paste in. ### Where to use it Any n8n that can reach Google and OpenAI will run it. The notes appear in the run, and a follow-up meeting lands in your Google Calendar. - n8n Cloud: import the file into a new workflow there - Your own n8n server: the same file and steps - Google Calendar: new follow-up events appear there with the attendees invited - Email or chat: add a node after \`AI Agent\` to send the summary to your team ### How to set it up Download saves a zip with `workflow.json` and Mindket’s setup notes in your downloads folder. Save the workflow once after you import it, because the agent’s scheduling tool calls the workflow by its own ID. What you need first | Need | Detail | | --- | --- | | n8n | On n8n Cloud or on a server of your own | | Google Meet | A call you organised, with transcription on, booked from a Google Calendar event | | Google Cloud | A project with the Google Meet REST API on, for a Google OAuth2 API credential | | Google | A Calendar credential and a Drive credential, each with OAuth2 | | OpenAI | An API key | 1. In n8n, create a workflow, open its menu, choose Import from File and pick `workflow.json`. Save it. 2. Add a Google Calendar OAuth2 credential. Select it in `Get Calendar Event`, `Create Calendar Event1` and `Add Attendee to Invite`. 3. In those three nodes, replace `` with your calendar’s ID. In `Get Calendar Event`, replace `abc123` with the ID of the calendar event for the call. 4. Add a Google OAuth2 API credential with the scope `https://www.googleapis.com/auth/meetings.space.readonly`. Select it in `Get Meeting ConferenceRecords` and `Get Meeting Transcript Location`. 5. Add a Google Drive OAuth2 credential and select it in `Get Transcript File`. 6. Add an OpenAI credential and select it in `OpenAI Chat Model1`. Open the node and pick a model. 7. Press Test workflow. ### How to use it Set the event ID in `Get Calendar Event`, then press Test workflow. The run reads the event, finds its call, fetches the transcript and asks the agent for notes. Open `AI Agent` after the run to read them. A follow-up meeting shows up in your calendar. What to change for a different result | To change | Where | | --- | --- | | What the agent writes | The numbered instructions in the prompt of \`AI Agent\` | | The answer’s fields | The JSON schema in \`Structured Output Parser\` | | The model | \`OpenAI Chat Model1\`, which sets none in the file | | Which meeting it reads | The event ID in \`Get Calendar Event\` | | What the agent may do | The allowed actions in the same prompt, and a route in \`Actions Router\` for each new tool | ### Example A project lead runs a weekly team sync that Google Meet transcribes. Near the end, two attendees say they should go through the budget again on Friday. The lead sets that week’s event ID and presses Test workflow. The agent answers with a short summary, a line or two for each attendee, the agreed next steps and the meeting it booked, with its title and invite link. The scheduling tool runs, and a Friday event appears in your calendar with both attendees on the invite. Read the notes against the call before you share them. A model can mishear a name or give a point to the wrong person. ### For newcomers and experts **If you are new.** n8n draws a workflow as boxes joined by lines, and each box, a node, does one job. The top row finds and reads the transcript, and the agent sits at its end. The bottom row is the agent’s scheduling tool, a short separate flow in the same file. Run it on a short test call before you trust it. **For experts.** The agent’s tool is this same workflow, called by its own ID with a `route` field, so `Actions Router` can hold more tools later. Where to look closer | Node | What it does | What to know | | --- | --- | --- | | \`Get Meeting ConferenceRecords\` | Lists conference records filtered by the event’s meeting code | A plain HTTP request with the Google OAuth2 API credential | | \`Get Transcript File\` | Downloads the transcript document as a PDF, with its ID read in ID mode | The original copy read it from the wrong place and in URL mode, and Mindket’s copy repairs both | | \`Schedule Meeting\` | A Workflow tool named \`create\_calendar\_event\` that runs this workflow with the route \`meetings.create\` | Its description holds the request schema the agent follows | | \`Structured Output Parser\` | Holds the answer to four fields | The schema writes \`items:\` with a colon for \`next\_steps\`, so that list’s items carry no declared type | ### Good to know Only a recorded call with a transcript works, and the agent can invite people on its own. - Transcripts: only a call you organised, with transcription on, has one. Without it the run stops at \`Get Transcript File\` - Scopes: Google’s Meet reference, read on 12 November 2026, lists \`meetings.space.created\` and \`meetings.space.readonly\` as the scopes for these calls - Repairs: the original copy read the transcript’s ID from the wrong place, gave it to Drive in URL mode and printed attendee names as undefined. Mindket’s copy fixes all three - Model: the chat model sets none, so n8n picks its default, \`gpt-5-mini\` in n8n’s source that day. OpenAI’s model page marks it Deprecated and points new work to \`gpt-5.6-terra\` - Shutdown: OpenAI’s deprecations page, read the same day, gives 14 January 2027 for \`gpt-4o\`, \`gpt-4o-mini\`, \`gpt-4.1\` and \`gpt-4.1-mini\`, with \`gpt-6-luna\` as the replacement - No review: the agent books and invites without waiting for you, and anything said in the call can steer it. Test with your own addresses first - Time zone: the agent writes the new event’s date and time as text. Check the first ones against your calendar - Billing: OpenAI bills each agent call, with the whole transcript in it - Your data: the transcript, the attendees’ names and email addresses and the start time go to OpenAI - Credentials: the JSON holds none ## README ```markdown # Google Meet transcript to notes and follow-up events An n8n workflow you start by hand. It takes one Google Calendar event, finds the Google Meet call attached to it, downloads the call's transcript from Google Drive and gives it to an OpenAI agent. The agent writes a summary, each person's main points and the next steps. If the attendees agreed to meet again, it also books that meeting in your calendar and invites them. ## What is inside - `workflow.json`: the workflow as n8n exports it, 19 nodes and 9 sticky notes. It holds no credential. - `README.md`: these notes. ## What we changed from the original copy - Three nodes pointed at the author's own calendar by its ID (`Get Calendar Event`, `Create Calendar Event1` and `Add Attendee to Invite`). Each now reads ``. - `Get Transcript File` read the transcript's address from `$json.docsDestination.document`. Google's reference for the Meet API says the transcript list answers with a `transcripts` array, and that `document` holds a bare document ID. The file had neither right. It now reads `$json.transcripts[0].docsDestination.document`, and the Drive node takes it in ID mode, because n8n's source makes URL mode accept only a full docs.google.com or drive.google.com address. Both were read on 12 November 2026. - The agent's prompt listed attendees by `attendee.display_name`. Google's calendar events carry `displayName`, so every name would have printed as undefined. It now reads `attendee.displayName`. ## Before you start - An n8n instance, on n8n Cloud or your own server. - A Google account that organised a Google Meet call with transcription turned on, and the Google Calendar event the call was booked from. - A Google Cloud project with the Google Meet REST API turned on, for a Google OAuth2 API credential. - An OpenAI API key. ## Set it up 1. In n8n, create a workflow, open its menu and choose Import from File. Pick `workflow.json`, then save it. The agent's scheduling tool calls this same workflow by its own ID, so it needs a saved copy. 2. Add a Google Calendar OAuth2 credential. Select it in `Get Calendar Event`, `Create Calendar Event1` and `Add Attendee to Invite`. 3. In those same three nodes, replace `` with the ID of the calendar the events live in. 4. In `Get Calendar Event`, replace `abc123` with the ID of the calendar event for the call. 5. Add a Google OAuth2 API credential with the scope `https://www.googleapis.com/auth/meetings.space.readonly`. Select it in `Get Meeting ConferenceRecords` and `Get Meeting Transcript Location`. 6. Add a Google Drive OAuth2 credential and select it in `Get Transcript File`. 7. Add an OpenAI credential and select it in `OpenAI Chat Model1`. Open the node and pick the model you want. 8. Press Test workflow. ## How a run goes 1. `When clicking "Test workflow"` starts the run by hand. 2. `Get Calendar Event` reads the event whose ID you set. 3. `Get Meeting ConferenceRecords` asks the Meet API for the conference record whose meeting code matches the event's `conferenceData.conferenceId`. 4. `Get Meeting Transcript Location` lists the transcripts of the first record. 5. `Get Transcript File` downloads the first transcript's Google Doc from Drive as a PDF, and `PDF Loader` pulls its text out. 6. `AI Agent` is an OpenAI Functions agent. Its prompt holds the meeting's creator, its attendees, its start time and the transcript. It asks for a summary with the meeting's goals, each attendee's main points and every next step, or up to three suggestions if none were agreed. It also asks for any follow-up calendar events. `Structured Output Parser` holds the answer to four fields: `summary`, `highlights`, `next_steps` and `meetings_created`. 7. When the attendees asked for another meeting, the agent calls `Schedule Meeting`, a tool named `create_calendar_event`. It runs this workflow again from `Execute Workflow Trigger` with the route `meetings.create`. 8. `Edit Fields` parses the agent's request, and `Actions Router` sends the `meetings.create` route to `Create Calendar Event1`. Any other route goes to `Fallback Response`, which says the tool was not found. 9. `Get Attendees` and `Attendees List` split the attendees, `Add Attendee to Invite` adds each one to the new event, and `Response` hands the event back to the agent as text. ## Change the look - What the agent writes: the numbered instructions in `AI Agent`'s prompt. - Which actions it may take: the same prompt says only calendar events are allowed. A new tool also needs its own route in `Actions Router`. - The shape of the answer: the JSON schema in `Structured Output Parser`. - The model: `OpenAI Chat Model1`. The file sets no model and a temperature of 0. - The meeting's source: `Get Calendar Event` fixes one event ID. Change it to read the latest event, or replace the node to pick a call another way. ## Good to know - Only a call you organised, with recording and transcription turned on, has a transcript. A call with none leaves `Get Transcript File` nothing to read, and the run stops there. - The model facts below come from pages read on 12 November 2026. - The Meet reference lists `meetings.space.created` and `meetings.space.readonly` as the scopes that open these two calls. The Drive download needs the Drive credential on top. - The chat model sets no model. n8n's source gives this node version `gpt-5-mini`, and OpenAI's model page marks that model Deprecated and points new work to `gpt-5.6-terra`. Pick a model in the node before you run it. - OpenAI's deprecations page gives 14 January 2027 as the shutdown date of `gpt-4o`, `gpt-4o-mini`, `gpt-4.1` and `gpt-4.1-mini`, and names `gpt-6-luna` as their replacement. - `AI Agent` is on version 1.5 with the OpenAI Functions agent type. In n8n's source, newer agent versions offer only the Tools agent. - The prompt asks the agent to book any follow-up meeting the attendees suggested, and nothing waits for your review. Whatever a person says in the call can shape what the agent books and who it invites. - The agent's instructions sit in the user message, since `options` leaves the system message at n8n's default. - Dates and times for the new event come from the agent as text. Check the first events it creates against your time zone. - Billing: OpenAI bills the agent's calls, each one holding the whole transcript. - The transcript, the attendees' names and email addresses and the event's start time go to OpenAI. Google handles the calendar, Meet and Drive requests. - The sticky notes keep the original's section titles and its links to n8n's documentation. ``` ## What is in it 3 files, 29 KB, downloaded as a zip. ```text NOTICE.md 650 B README.md 7 KB workflow.json 22 KB ``` ## Workflow check Checked by Mindket on 13 Nov 2026. Read from the workflow’s file, not from a run. - Steps: 16 steps, starting when another workflow calls it or by hand - AI: 1 step calls an AI model: AI Agent - Sign-ins: Signs in to: Google Calendar, Google, Google Drive and OpenAI - API keys: No API key is left in the file - Step types: Mindket’s checker reads all 10 of its step types ## Get it - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/b3ee619d-a42a-499b-966d-5bb8284f27ab) - [Its page on Mindket](https://mindket.com/p/b3ee619d-a42a-499b-966d-5bb8284f27ab)