Fireflies.ai Data Handling

Overview

Manage meeting transcript data from Fireflies.ai. Covers transcript export formats, PII redaction in transcripts, meeting data retention policies, and selective data sync to CRM and project management tools.

Prerequisites

• Fireflies.ai API key
• GraphQL client configured
• Understanding of transcript data structure
• Storage for processed transcripts

Instructions

Step 1: Transcript Export and Format Conversion

import { GraphQLClient } from 'graphql-request';

const fireflies = new GraphQLClient('https://api.fireflies.ai/graphql', {
  headers: { Authorization: `Bearer ${process.env.FIREFLIES_API_KEY}` },
});

const FULL_TRANSCRIPT = `
  query GetTranscript($id: String!) {
    transcript(id: $id) {
      id title date duration
      sentences { speaker_name text start_time end_time }
      summary { overview action_items keywords }
    }
  }
`;

async function exportTranscript(id: string, format: 'json' | 'text' | 'srt') {
  const { transcript } = await fireflies.request(FULL_TRANSCRIPT, { id });

  switch (format) {
    case 'json':
      return JSON.stringify(transcript, null, 2);
    case 'text':
      return transcript.sentences
        .map((s: any) => `${s.speaker_name}: ${s.text}`)
        .join('\n');
    case 'srt':
      return transcript.sentences
        .map((s: any, i: number) => [
          i + 1,
          `${formatTime(s.start_time)} --> ${formatTime(s.end_time)}`,
          `${s.speaker_name}: ${s.text}`,
          '',
        ].join('\n'))
        .join('\n');
  }
}

function formatTime(seconds: number): string {
  const h = Math.floor(seconds / 3600);  # 3600: timeout: 1 hour
  const m = Math.floor((seconds % 3600) / 60);  # timeout: 1 hour
  const s = Math.floor(seconds % 60);
  const ms = Math.floor((seconds % 1) * 1000);  # 1000: 1 second in ms
  return `${String(h).padStart(2,'0')}:${String(m).padStart(2,'0')}:${String(s).padStart(2,'0')},${String(ms).padStart(3,'0')}`;
}

Step 2: PII Redaction in Transcripts

const PII_PATTERNS = [
  { regex: /\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,}\b/gi, tag: '[EMAIL]' },
  { regex: /\b\d{3}[-.]?\d{3}[-.]?\d{4}\b/g, tag: '[PHONE]' },
  { regex: /\b\d{3}-\d{2}-\d{4}\b/g, tag: '[SSN]' },
  { regex: /\b\d{4}[\s-]?\d{4}[\s-]?\d{4}[\s-]?\d{4}\b/g, tag: '[CARD]' },
];

function redactTranscript(sentences: any[]) {
  return sentences.map(s => ({
    ...s,
    text: redactText(s.text),
  }));
}

function redactText(text: string): string {
  let redacted = text;
  for (const { regex, tag } of PII_PATTERNS) {
    redacted = redacted.replace(regex, tag);
  }
  return redacted;
}

Step 3: Meeting Data Retention Policy

interface RetentionPolicy {
  transcriptRetentionDays: number;
  summaryRetentionDays: number;
  actionItemRetentionDays: number;
}

const DEFAULT_POLICY: RetentionPolicy = {
  transcriptRetentionDays: 90,
  summaryRetentionDays: 365,  # 365 days = 1 year
  actionItemRetentionDays: 180,
};

async function applyRetention(
  transcripts: any[],
  policy = DEFAULT_POLICY
) {
  const now = Date.now();
  const results = { kept: 0, archived: 0, deleted: 0 };

  for (const t of transcripts) {
    const ageDays = (now - new Date(t.date).getTime()) / 86400000;  # 86400000 = configured value

    if (ageDays > policy.transcriptRetentionDays) {
      // Archive: keep summary, delete full transcript
      await archiveTranscript(t.id, {
        keepSummary: ageDays <= policy.summaryRetentionDays,
        keepActions: ageDays <= policy.actionItemRetentionDays,
      });
      results.archived++;
    } else {
      results.kept++;
    }
  }

  return results;
}

Step 4: Selective CRM Sync

async function syncActionItemsToCRM(transcriptId: string) {
  const { transcript } = await fireflies.request(FULL_TRANSCRIPT, { id: transcriptId });

  const actionItems = transcript.summary?.action_items || [];
  if (actionItems.length === 0) return { synced: 0 };

  const tasks = actionItems.map((item: string) => ({
    title: item.slice(0, 200),  # HTTP 200 OK
    source: 'fireflies',
    meetingTitle: transcript.title,
    meetingDate: transcript.date,
    participants: transcript.sentences
      .map((s: any) => s.speaker_name)
      .filter((v: string, i: number, a: string[]) => a.indexOf(v) === i),
  }));

  return { synced: tasks.length, tasks };
}

Error Handling

Issue	Cause	Solution
Missing sentences	Transcription still processing	Check transcript status before export
PII in action items	Redaction only applied to sentences	Also redact summary fields
Large transcript	Long meeting (2+ hours)	Process in chunks, stream export
Retention policy gap	No automated cleanup	Schedule weekly retention job

Examples

Batch Export for Compliance

async function exportAllForAudit(ids: string[]) {
  return Promise.all(ids.map(async id => ({
    id,
    text: await exportTranscript(id, 'text'),
    exportedAt: new Date().toISOString(),
  })));
}

Resources

• Fireflies GraphQL API
• Fireflies Data Retention

Output

• Configuration files or code changes applied to the project
• Validation report confirming correct implementation
• Summary of changes made and their rationale