Automated Meeting Summaries from Fireflies Transcripts using AI - n8n Workflow

Business Analysts & Project Managers: Who need quick, reliable summaries of client or internal meetings. Automation Enthusiasts: Seeking advanced examples of using the n8n LangChain nodes with powerful LLMs like Google Gemini. Sales Teams: Who rely on accurate meeting recaps for follow-ups and CRM updates. Anyone utilizing Fireflies.ai for meeting transcription and wanting a custom delivery mechanism via a robust n8n workflow.

This sophisticated n8n workflow addresses the challenge of managing and disseminating information captured in long meeting transcripts. By leveraging the power of Fireflies.ai and integrating it with Google Gemini via the LangChain n8n node, this solution transforms raw transcription data into structured, actionable summaries and professional client emails. The n8n trigger for this automation can be a Fireflies webhook, a manual execution, or, most commonly, a polling Gmail trigger that detects the completion of a meeting recap, ensuring the automation is highly responsive. This setup is a prime example of utilizing flexible n8n templates to build complex, multi-stage AI processes, delivering immediate value by saving hours on manual recap creation.

This n8n workflow provides three distinct starting points (n8n trigger mechanisms) to initiate the process:

1. Gmail Trigger Path: The automation starts with the Gmail Trigger n8n node, polling for emails from [email protected] with the subject 'Your meeting recap'.


2. ID Extraction: A Gmail node fetches the full message content. A subsequent Set n8n node prepares the data, which is then passed to an Information Extractor (using the OpenAI n8n node) to isolate the Fireflies meeting link.


3. Meeting ID Resolution: A Code n8n node processes the extracted URL string to cleanly isolate the meetingId required for the Fireflies API call.


4. Transcript Retrieval: The Fireflies 'Get a transcript' n8n node uses the extracted meetingId to fetch the complete transcription data, including speaker segmentation and initial summaries provided by Fireflies.


5. Data Split and Summarization: The flow splits into two primary AI branches:

1. Client Email Drafting: Concurrently, the summaries from both AI paths are merged, aggregated, and fed into the 'Email writer' (another Google Gemini n8n node). This AI is instructed to combine the information into a professional client-facing email draft, which the final Gmail n8n node saves as a draft, ready for quick review and sending.

1. Import the n8n workflow: Copy the provided JSON data and paste it into your n8n instance under the 'Workflows' section using the 'New' -> 'Import from JSON' option.


2. Credentials Setup: You will need to configure three primary sets of credentials:

1. Configuration: Update placeholder values in the Gmail n8n nodes, specifically changing YOUR_EMAIL to the recipient address for the summaries.


2. Activation: Once credentials are set for every n8n node, activate the workflow. For the Gmail trigger path, the workflow will run based on the defined polling interval. For webhook use, configure the n8n webhook URL in your Fireflies settings.

Gmail Trigger (n8n trigger): Sets the automation in motion by watching for Fireflies recap emails (subject:Your meeting recap, sender:[email protected]). It uses a specific query filter to ensure targeted polling.
Get a message (Gmail n8n node): Retrieves the full content of the email identified by the trigger, necessary for link extraction.
Information Extractor (LangChain n8n node - OpenAI): This node uses an LLM (configured here as gpt-4.1-mini) with a system prompt to act as an expert extraction algorithm. Its goal is strictly to extract the meeting_notes link from the email body.
Code1 (Code n8n node): Essential for transforming the extracted URL string. This JavaScript snippet finds the meeting ID by slicing the string between '::' and '?'—a critical step before hitting the Fireflies API.
Get a transcript (Fireflies n8n node): The core integration point. It takes the dynamically generated meetingId and fetches all transcript data from Fireflies.
Full transcript (Code n8n node): Reconstructs the full, chronological conversation history by looping through the sentences array from the Fireflies data and combining speaker names and dialogue. This raw input is vital for the AI.
Expert Meeting transcripts & Meeting summary expert (Google Gemini n8n node): These are the powerful AI processing centers. They use the models/gemini-2.5-pro model with system prompts instructing them to act as expert interpreters and summarizers, generating the final recap content.
Merge & Aggregate (Core Logic n8n nodes): Used to combine and structure the outputs from the separate AI summarization paths before feeding them into the final Email writer node.

• Draft email to client (Gmail n8n node): The final action, which saves the professionally written AI draft email in the user's Gmail account. This flexible n8n node handles both sending and drafting operations.