How to Transcribe Google Meet Recordings with Jinba

Summary

Jinba has no direct Google Meet connector, so the workflow is file-based: INPUT_FILE ingests the exported transcript, VTT_PARSE handles .vtt files, and an AI invocation tool generates the summary.
Paid Google Workspace plans can export a Google Doc transcript as plain text; free or Business Starter plans need a speech-to-text service to produce a WebVTT .vtt file.
For .vtt files, set VTT_PARSE to txt to strip timestamps before summarization; for plain text exports, pass the text directly to the AI step.
Once the transcript is hosted at a predictable URL, the three-step recipe parses, summarizes, and archives the meeting record automatically. Build this workflow in Jinba Flow.

Jinba does not include a GOOGLE_MEET_* connector. There is no tool in the Jinba catalog that pulls directly from a Google Meet session. What Jinba does include is a file-based workflow, built on INPUT_FILE, VTT_PARSE, and an AI invocation tool, that turns an exported transcript into a structured, high-quality summary. The handoff step is manual: you export the file from Google Meet, then Jinba takes over from there.

This distinction matters because it shapes how to build the workflow. The steps below cover both paths to an exportable file, one for paid Google Workspace users and one for those on free or Business Starter plans, then walk through the full Jinba recipe from file ingestion to archiving.

Why there is no direct Google Meet integration

Google Meet's native transcription does not expose a clean developer API for third-party tools to consume, as documented by Nylas. Programmatic access to Meet transcripts requires working through the Google Meet REST API's conferenceRecords route, which adds significant complexity and is outside the scope of a no-code workflow tool. This is not a gap specific to Jinba; it reflects the state of Google Meet's developer surface.

What this means in practice: any robust google meet automation strategy that avoids bespoke API development starts with an export step. The transcript leaves Google Meet as a file, and that file becomes the input to everything downstream.

Step 1: Export your transcript from Google Meet

The export path depends on your Google Workspace plan.

Paid Google Workspace plans (Business Standard, Business Plus, Enterprise)

For most paid editions, Google Meet generates a transcript automatically and saves it as a Google Doc in the meeting organiser's Drive under a folder called "Meet Recordings." The document includes speaker labels and timestamps. Chat messages are not included.

To use this transcript in Jinba:

Open the Google Doc from Drive.
Export it as a plain text file: File > Download > Plain Text (.txt).
Host the file at a publicly accessible URL, such as a Google Drive share link set to "Anyone with the link can view," or upload it to any file host that returns a direct download URL.

Because the Google Doc export is plain text rather than WebVTT format, VTT_PARSE will not apply directly. Use the plain text block as the input to your AI summarisation step instead.

Free and Business Starter plans

These plans do not include native transcription. The practical path is to record the meeting as an audio or video file, then run it through a speech-to-text service that outputs a WebVTT (.vtt) file. Services such as Google Cloud Speech-to-Text and similar transcription APIs support .vtt output. Once you have the .vtt file, host it at an accessible URL and proceed to Step 2.

Step 2: Process the transcript in Jinba

2.1 Ingest the file with `INPUT_FILE`

The INPUT_FILE tool is the entry point for any file-based workflow in Jinba. It accepts a URL pointing to your transcript file and makes the file available to every subsequent step.

Add an INPUT_FILE step named upload_transcript and set its url input to the direct download link for your .vtt or .txt file. The result of this step, {{steps.upload_transcript.result.url}}, is what you will pass into the next tool.

2.2 Parse the transcript with `VTT_PARSE`

VTT_PARSE is Jinba's dedicated tool for processing WebVTT files. It accepts two inputs:

file_url: the URL of the .vtt file, passed from the previous step as {{steps.upload_transcript.result.url}}
output_format: either txt or array

The choice of output format determines what the next step receives:

txt strips all timestamps and formatting markup and returns a single block of clean speech text at result.text. This is the correct choice when the next step is an AI summarisation tool.
array returns a structured list of cues at result.cues. Each cue contains an index, start and end timestamps in HH:MM:SS.mmm format, and the cue text. Use this format when you need time-anchored notes, searchable segments, or downstream analysis at the cue level.

For summarisation, set output_format to txt.

If the file URL is inaccessible, VTT_PARSE returns a DownloadError. If the file content is not valid WebVTT, it returns a ParseError. Both errors surface in the workflow run log without requiring additional error-handling configuration.

2.3 Generate a structured summary with `OPENAI_INVOKE` or `ANTHROPIC_INVOKE`

A recurring issue with existing meeting assistant tools is summaries that are short and generic rather than actionable. The quality of the output depends almost entirely on the structure of the prompt. The following prompt, drawn from Jinba's "Meeting Transcript to Summary" recipe, produces summaries with the specificity teams actually need:

Based on the following meeting transcript, please provide a summary that includes:
1. Key discussion points.
2. All decisions that were made.
3. A list of action items, including who is assigned to each item.

Transcript:
{{steps.parse_transcript.result.text}}

Pass this prompt to either OPENAI_INVOKE (using gpt-4 or a later model) or ANTHROPIC_INVOKE (using Claude). The Jinba tool catalog also includes connectors for Azure OpenAI, Gemini, Grok, and LlamaCloud, which allows enterprise teams to route inference through whichever model aligns with their governance or data residency requirements.

Step 3: Archive the output

Summarisation is not the final step. Teams that stop there still face the manual task of copying the output somewhere accessible. Jinba's integration tools complete the loop.

Common archiving patterns:

Google Sheets: Use the "Add a Row" action to append the meeting date, a link to the recording, and the AI-generated summary to a shared meeting log.
Notion: Write the summary directly to a Notion database, with the meeting date and attendees as properties.
Microsoft Teams or SharePoint: Post the summary to a Teams channel or store it in a SharePoint document library.

All of these integrations are available in the Jinba tool catalog.

The complete Jinba recipe

The three steps below form a complete, copy-pasteable workflow. Name each step as shown so that the variable references resolve correctly.

# Step 1: Ingest the transcript file
name: upload_transcript
tool: INPUT_FILE
inputs:
url: "YOUR_TRANSCRIPT_FILE_URL"

# Step 2: Parse the VTT file into clean text
name: parse_transcript
tool: VTT_PARSE
inputs:
file_url: "{{steps.upload_transcript.result.url}}"
output_format: txt

# Step 3: Generate a structured summary
name: summarize_transcript
tool: OPENAI_INVOKE
inputs:
model: gpt-4
prompt: |
Based on the following meeting transcript, please provide a summary that includes:
1. Key discussion points.
2. All decisions that were made.
3. A list of action items, including who is assigned to each item.

Transcript:
{{steps.parse_transcript.result.text}}

Replace OPENAI_INVOKE with ANTHROPIC_INVOKE and update the model value if your team uses Claude. Add a fourth step using the Google Sheets or Notion tool to send {{steps.summarize_transcript.result.text}} to your preferred archive destination.

What to do next

The export step is the one part of this workflow that remains manual. For teams running frequent meetings, that friction adds up. A practical next step is to establish a consistent file-hosting convention, a shared Drive folder, a designated S3 bucket, or a simple upload endpoint, so that the URL passed to INPUT_FILE is always in a predictable location.

Once the file location is standardised, the rest of the workflow runs without intervention. The transcript is parsed, the summary is generated with the structure your team needs, and the output is archived where the team can find it. That covers the full cycle from exported Google Meet file to actionable meeting record, without a direct integration and without the abbreviated summaries that make other meeting assistant tools a poor fit for substantive discussions.

Build the workflow in Jinba and run it against your next meeting export to verify the output quality before committing to a broader rollout.

Frequently Asked Questions

Does Jinba integrate directly with Google Meet?

No. Jinba does not include a GOOGLE_MEET_* connector. Instead, it provides a file-based workflow that uses INPUT_FILE, VTT_PARSE, and an AI invocation tool to convert an exported Google Meet transcript into a structured summary. This means you export the transcript from Google Meet first, then Jinba automates the parsing, summarization, and archiving steps.

How do I export a Google Meet transcript for Jinba?

The export path depends on your Google Workspace plan. If you are on a paid plan such as Business Standard, Business Plus, or Enterprise, Google Meet usually saves the transcript as a Google Doc in the meeting organizer’s Drive under “Meet Recordings.” Open that Doc and download it as a plain text file via File > Download > Plain Text (.txt). If you are on a free or Business Starter plan, record the meeting and use a speech-to-text service that outputs a WebVTT .vtt file, then host that file at a publicly accessible URL.

Can I use Jinba with a free Google Meet account?

Yes, you can still use Jinba, but the workflow requires an extra transcription step because free and Business Starter plans do not include native Google Meet transcription. Record the meeting as an audio or video file, run it through a transcription service that produces a .vtt file, and then feed that file into Jinba. Once the .vtt file is hosted at an accessible URL, the rest of the workflow is the same as for paid plans.

What file formats does Jinba support for Google Meet transcripts?

Jinba works with two main transcript file types: WebVTT .vtt files and plain text .txt files. Use VTT_PARSE for .vtt files and set the output format to txt for clean transcript text. If you export a Google Doc as a plain text .txt file, skip VTT_PARSE and pass that text directly to your AI summarization step.

Why doesn’t Jinba connect directly to Google Meet?

Google Meet’s native transcription does not expose a clean developer API for third-party tools to consume. Programmatic access requires working through the Google Meet REST API’s conferenceRecords route, which adds significant complexity and is outside the scope of a no-code workflow tool. This is not specific to Jinba; any robust Google Meet automation strategy that avoids bespoke API development typically starts with an export step.

How do I fix a `VTT_PARSE` error in Jinba?

A DownloadError means the file URL is inaccessible. Check that the URL is publicly reachable and returns a direct download rather than an HTML preview page. A ParseError means the file content is not valid WebVTT. In that case, verify that the file was exported as a true .vtt file and is not a Google Doc plain text export, which should be passed directly to the AI step instead.

Which AI models can I use to summarize Google Meet transcripts in Jinba?

You can use OPENAI_INVOKE with gpt-4 or a later model, or ANTHROPIC_INVOKE with Claude. Jinba also includes connectors for Azure OpenAI, Gemini, Grok, and LlamaCloud, which allows enterprise teams to route inference through whichever model aligns with their governance or data residency requirements. For best results, use the structured prompt from the recipe covering key discussion points, decisions, and action items with assignees.

How do I archive Google Meet summaries after Jinba processes them?

Jinba supports several archiving integrations. You can append the meeting date, recording link, and AI-generated summary to a Google Sheet using the “Add a Row” action, write the summary to a Notion database, or post it to a Microsoft Teams channel or SharePoint document library. Add a fourth step to the workflow that sends {{steps.summarize_transcript.result.text}} to your preferred destination.

Can I fully automate Google Meet summaries with Jinba?

Not entirely end to end. The export step remains manual because you must export the transcript from Google Meet and host it at a URL that INPUT_FILE can reach. However, once that file location is standardized, the rest of the workflow—parsing, summarization, and archiving—runs automatically without further intervention.

Where can I find the Jinba workflow template for Google Meet transcript summarization?

The complete recipe is included earlier in this article. It consists of three steps: upload_transcript with INPUT_FILE, parse_transcript with VTT_PARSE, and summarize_transcript with OPENAI_INVOKE or ANTHROPIC_INVOKE. Copy the YAML, replace the file URL and model value, and add an optional fourth step for archiving.