Twinmind Migration Deep Dive

TwinMind Migration Deep Dive

Overview

Comprehensive guide for migrating from other meeting AI services to TwinMind.

Prerequisites

• TwinMind Pro/Enterprise account
• Export access on source platform
• Understanding of data formats
• Sufficient storage for migration

Supported Migration Sources

Instructions

Step 1: Assess Current Data

// scripts/migration-assessment.ts
import * as fs from 'fs';
import * as path from 'path';

interface MigrationAssessment {
  source: string;
  totalTranscripts: number;
  totalDurationHours: number;
  dateRange: { earliest: Date; latest: Date };
  dataFormats: string[];
  estimatedMigrationTime: string;
  potentialIssues: string[];
}

async function assessMigration(
  source: 'otter' | 'fireflies' | 'rev' | 'zoom' | 'teams',
  dataPath: string
): Promise<MigrationAssessment> {
  const files = fs.readdirSync(dataPath);
  const issues: string[] = [];

  let totalTranscripts = 0;
  let totalDuration = 0;
  let earliest = new Date();
  let latest = new Date(0);
  const formats = new Set<string>();

  for (const file of files) {
    const ext = path.extname(file).toLowerCase();
    formats.add(ext);

    const filePath = path.join(dataPath, file);
    const stat = fs.statSync(filePath);

    if (ext === '.json') {
      try {
        const content = JSON.parse(fs.readFileSync(filePath, 'utf-8'));
        totalTranscripts++;

        // Extract duration if available
        if (content.duration) {
          totalDuration += content.duration;
        }

        // Extract date
        const date = new Date(content.created_at || content.date || stat.mtime);
        if (date < earliest) earliest = date;
        if (date > latest) latest = date;
      } catch (e) {
        issues.push(`Failed to parse: ${file}`);
      }
    }
  }

  // Estimate migration time (rough: 1 transcript per second + processing)
  const estimatedMinutes = Math.ceil(totalTranscripts / 60) + Math.ceil(totalDuration / 3600 * 10);

  return {
    source,
    totalTranscripts,
    totalDurationHours: totalDuration / 3600,
    dateRange: { earliest, latest },
    dataFormats: Array.from(formats),
    estimatedMigrationTime: `${estimatedMinutes} minutes`,
    potentialIssues: issues,
  };
}

// Run assessment
const assessment = await assessMigration('otter', './exports/otter');
console.log('Migration Assessment:');
console.log(JSON.stringify(assessment, null, 2));

Step 2: Export Data from Source

Otter.ai Export

// scripts/exporters/otter.ts
import puppeteer from 'puppeteer';

async function exportFromOtter(credentials: { email: string; password: string }) {
  const browser = await puppeteer.launch({ headless: false });
  const page = await browser.newPage();

  // Login
  await page.goto('https://otter.ai/login');
  await page.type('input[name="email"]', credentials.email);
  await page.type('input[name="password"]', credentials.password);
  await page.click('button[type="submit"]');
  await page.waitForNavigation();

  // Navigate to conversations
  await page.goto('https://otter.ai/my-notes');
  await page.waitForSelector('.conversation-list');

  // Get all conversation IDs
  const conversationIds = await page.$$eval(
    '.conversation-item',
    items => items.map(item => item.getAttribute('data-id'))
  );

  console.log(`Found ${conversationIds.length} conversations`);

  // Export each conversation
  const exports = [];
  for (const id of conversationIds) {
    const data = await page.evaluate(async (convId) => {
      const response = await fetch(`/api/conversation/${convId}/export`);
      return response.json();
    }, id);

    exports.push(data);
  }

  await browser.close();
  return exports;
}

Fireflies.ai Export

// scripts/exporters/fireflies.ts
const FIREFLIES_API_URL = 'https://api.fireflies.ai/graphql';

async function exportFromFireflies(apiKey: string): Promise<any[]> {
  const query = `
    query GetAllTranscripts {
      transcripts {
        id
        title
        date
        duration
        transcript_text
        summary
        action_items {
          text
          assignee
        }
        speakers {
          name
          id
        }
        sentences {
          text
          speaker_id
          start_time
          end_time
        }
      }
    }
  `;

  const response = await fetch(FIREFLIES_API_URL, {
    method: 'POST',
    headers: {
      'Authorization': `Bearer ${apiKey}`,
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({ query }),
  });

  const data = await response.json();
  return data.data.transcripts;
}

Rev.ai Export

// scripts/exporters/rev.ts
async function exportFromRev(apiKey: string): Promise<any[]> {
  // Get all jobs
  const jobsResponse = await fetch('https://api.rev.ai/speechtotext/v1/jobs', {
    headers: { 'Authorization': `Bearer ${apiKey}` },
  });

  const jobs = await jobsResponse.json();
  const transcripts = [];

  for (const job of jobs) {
    if (job.status === 'transcribed') {
      // Get transcript
      const transcriptResponse = await fetch(
        `https://api.rev.ai/speechtotext/v1/jobs/${job.id}/transcript`,
        {
          headers: {
            'Authorization': `Bearer ${apiKey}`,
            'Accept': 'application/vnd.rev.transcript.v1.0+json',
          },
        }
      );

      const transcript = await transcriptResponse.json();
      transcripts.push({
        id: job.id,
        name: job.name,
        created_on: job.created_on,
        duration_seconds: job.duration_seconds,
        transcript,
      });
    }
  }

  return transcripts;
}

Step 3: Transform Data to TwinMind Format

// scripts/migration/transform.ts
import { Transcript, Segment, Speaker } from '../../src/twinmind/types';

interface TransformOptions {
  preserveTimestamps: boolean;
  preserveSpeakers: boolean;
  generateNewIds: boolean;
}

// Otter.ai to TwinMind
export function transformOtterToTwinMind(otterData: any): Transcript {
  return {
    id: `tm_imported_${otterData.id}`,
    text: otterData.text || otterData.transcript_text,
    duration_seconds: otterData.duration || 0,
    language: otterData.language || 'en',
    created_at: otterData.created_at || new Date().toISOString(),
    metadata: {
      imported_from: 'otter.ai',
      original_id: otterData.id,
    },
    speakers: otterData.speakers?.map((s: any, i: number) => ({
      id: `spk_${i}`,
      name: s.name || s.speaker_name || `Speaker ${i + 1}`,
    })),
    segments: otterData.segments?.map((seg: any) => ({
      start: seg.start_time || seg.start,
      end: seg.end_time || seg.end,
      text: seg.text || seg.transcript,
      speaker_id: seg.speaker_id,
      confidence: seg.confidence || 0.95,
    })),
  };
}

// Fireflies.ai to TwinMind
export function transformFirefliesToTwinMind(ffData: any): Transcript {
  return {
    id: `tm_imported_${ffData.id}`,
    title: ffData.title,
    text: ffData.transcript_text,
    duration_seconds: ffData.duration,
    language: 'en',  // Fireflies doesn't export language
    created_at: ffData.date,
    metadata: {
      imported_from: 'fireflies.ai',
      original_id: ffData.id,
      summary: ffData.summary,
      action_items: ffData.action_items,
    },
    speakers: ffData.speakers?.map((s: any) => ({
      id: s.id,
      name: s.name,
    })),
    segments: ffData.sentences?.map((sent: any) => ({
      start: sent.start_time,
      end: sent.end_time,
      text: sent.text,
      speaker_id: sent.speaker_id,
      confidence: 0.95,
    })),
  };
}

// Rev.ai to TwinMind
export function transformRevToTwinMind(revData: any): Transcript {
  const monologues = revData.transcript.monologues || [];

  const segments: Segment[] = [];
  const speakers = new Map<number, string>();

  for (const monologue of monologues) {
    const speakerId = monologue.speaker;
    if (!speakers.has(speakerId)) {
      speakers.set(speakerId, `Speaker ${speakerId + 1}`);
    }

    for (const element of monologue.elements) {
      if (element.type === 'text') {
        segments.push({
          start: element.ts,
          end: element.end_ts,
          text: element.value,
          speaker_id: `spk_${speakerId}`,
          confidence: element.confidence || 0.95,
        });
      }
    }
  }

  return {
    id: `tm_imported_${revData.id}`,
    title: revData.name,
    text: segments.map(s => s.text).join(' '),
    duration_seconds: revData.duration_seconds,
    language: 'en',
    created_at: revData.created_on,
    metadata: {
      imported_from: 'rev.ai',
      original_id: revData.id,
    },
    speakers: Array.from(speakers.entries()).map(([id, name]) => ({
      id: `spk_${id}`,
      name,
    })),
    segments,
  };
}

// Zoom VTT to TwinMind
export function transformZoomVTTToTwinMind(
  vttContent: string,
  metadata: { title: string; date: string; duration: number }
): Transcript {
  const segments: Segment[] = [];
  const lines = vttContent.split('\n');

  let currentSegment: Partial<Segment> = {};

  for (let i = 0; i < lines.length; i++) {
    const line = lines[i].trim();

    // Parse timestamp line: 00:00:00.000 --> 00:00:02.000
    if (line.includes('-->')) {
      const [start, end] = line.split('-->').map(t => parseVTTTimestamp(t.trim()));
      currentSegment.start = start;
      currentSegment.end = end;
    }
    // Parse speaker and text
    else if (line && !line.match(/^\d+$/) && currentSegment.start !== undefined) {
      // Format: "Speaker Name: Text" or just "Text"
      const speakerMatch = line.match(/^(.+?):\s*(.*)$/);

      if (speakerMatch) {
        currentSegment.speaker_id = speakerMatch[1];
        currentSegment.text = speakerMatch[2];
      } else {
        currentSegment.text = line;
      }

      if (currentSegment.text) {
        segments.push({
          start: currentSegment.start,
          end: currentSegment.end || currentSegment.start + 1,
          text: currentSegment.text,
          speaker_id: currentSegment.speaker_id,
          confidence: 0.9,
        } as Segment);
        currentSegment = {};
      }
    }
  }

  // Extract unique speakers
  const speakerIds = [...new Set(segments.map(s => s.speaker_id).filter(Boolean))];

  return {
    id: `tm_imported_zoom_${Date.now()}`,
    title: metadata.title,
    text: segments.map(s => s.text).join(' '),
    duration_seconds: metadata.duration,
    language: 'en',
    created_at: metadata.date,
    metadata: {
      imported_from: 'zoom',
    },
    speakers: speakerIds.map((id, i) => ({ id: `spk_${i}`, name: id as string })),
    segments,
  };
}

function parseVTTTimestamp(timestamp: string): number {
  const parts = timestamp.split(':');
  const seconds = parts.pop()!;
  const minutes = parts.pop() || '0';
  const hours = parts.pop() || '0';

  return (
    parseFloat(hours) * 3600 +
    parseFloat(minutes) * 60 +
    parseFloat(seconds)
  );
}

Step 4: Import to TwinMind

// scripts/migration/import.ts
import { getTwinMindClient } from '../../src/twinmind/client';
import { Transcript } from '../../src/twinmind/types';

interface ImportResult {
  successful: number;
  failed: number;
  errors: Array<{ id: string; error: string }>;
}

export async function importToTwinMind(
  transcripts: Transcript[],
  options: {
    batchSize?: number;
    delayMs?: number;
    dryRun?: boolean;
  } = {}
): Promise<ImportResult> {
  const client = getTwinMindClient();
  const batchSize = options.batchSize || 10;
  const delayMs = options.delayMs || 1000;

  const result: ImportResult = {
    successful: 0,
    failed: 0,
    errors: [],
  };

  console.log(`Starting import of ${transcripts.length} transcripts...`);

  for (let i = 0; i < transcripts.length; i += batchSize) {
    const batch = transcripts.slice(i, i + batchSize);

    for (const transcript of batch) {
      try {
        if (options.dryRun) {
          console.log(`[DRY RUN] Would import: ${transcript.title || transcript.id}`);
          result.successful++;
          continue;
        }

        // Check if already imported
        const existing = await client.get('/transcripts', {
          params: {
            'metadata.original_id': transcript.metadata?.original_id,
          },
        });

        if (existing.data.length > 0) {
          console.log(`Skipping duplicate: ${transcript.title || transcript.id}`);
          continue;
        }

        // Import transcript
        await client.post('/transcripts/import', {
          text: transcript.text,
          title: transcript.title,
          duration_seconds: transcript.duration_seconds,
          language: transcript.language,
          created_at: transcript.created_at,
          metadata: transcript.metadata,
          speakers: transcript.speakers,
          segments: transcript.segments,
        });

        result.successful++;
        console.log(`Imported: ${transcript.title || transcript.id}`);
      } catch (error: any) {
        result.failed++;
        result.errors.push({
          id: transcript.id,
          error: error.message,
        });
        console.error(`Failed: ${transcript.id} - ${error.message}`);
      }
    }

    // Rate limit delay between batches
    if (i + batchSize < transcripts.length) {
      await new Promise(r => setTimeout(r, delayMs));
    }
  }

  return result;
}

Step 5: Migration Verification

// scripts/migration/verify.ts
interface VerificationResult {
  sourceCount: number;
  importedCount: number;
  matchRate: number;
  missingTranscripts: string[];
  contentMismatches: Array<{
    id: string;
    issue: string;
  }>;
}

export async function verifyMigration(
  sourceTranscripts: Transcript[],
  importedIds: string[]
): Promise<VerificationResult> {
  const client = getTwinMindClient();
  const missingTranscripts: string[] = [];
  const contentMismatches: Array<{ id: string; issue: string }> = [];

  for (const source of sourceTranscripts) {
    // Find in TwinMind
    const imported = await client.get('/transcripts', {
      params: {
        'metadata.original_id': source.metadata?.original_id,
      },
    });

    if (imported.data.length === 0) {
      missingTranscripts.push(source.id);
      continue;
    }

    const importedTranscript = imported.data[0];

    // Verify content
    if (importedTranscript.text !== source.text) {
      const sourceWords = source.text.split(/\s+/).length;
      const importedWords = importedTranscript.text.split(/\s+/).length;

      if (Math.abs(sourceWords - importedWords) > sourceWords * 0.1) {
        contentMismatches.push({
          id: source.id,
          issue: `Word count mismatch: ${sourceWords} vs ${importedWords}`,
        });
      }
    }

    // Verify duration
    if (Math.abs(importedTranscript.duration_seconds - source.duration_seconds) > 1) {
      contentMismatches.push({
        id: source.id,
        issue: `Duration mismatch: ${source.duration_seconds}s vs ${importedTranscript.duration_seconds}s`,
      });
    }
  }

  return {
    sourceCount: sourceTranscripts.length,
    importedCount: sourceTranscripts.length - missingTranscripts.length,
    matchRate: (sourceTranscripts.length - missingTranscripts.length) / sourceTranscripts.length,
    missingTranscripts,
    contentMismatches,
  };
}

Step 6: Post-Migration Checklist

## Post-Migration Checklist

### Data Verification
- [ ] All transcripts imported (compare counts)
- [ ] Spot-check 10 random transcripts for content accuracy
- [ ] Verify speaker labels preserved
- [ ] Verify timestamps are accurate
- [ ] Check metadata (dates, durations) are correct

### Feature Verification
- [ ] Search finds imported transcripts
- [ ] Summaries can be generated for imported data
- [ ] Action items can be extracted
- [ ] Calendar integration works with imported meetings

### User Acceptance
- [ ] Key users can find their transcripts
- [ ] Historical data accessible
- [ ] Reports include migrated data

### Cleanup
- [ ] Remove temporary import files
- [ ] Update user documentation
- [ ] Archive source platform export
- [ ] Cancel source platform subscription

### Performance
- [ ] Search performance acceptable with migrated data
- [ ] No increase in error rates
- [ ] Memory usage within limits

Output

• Migration assessment tool
• Platform-specific exporters
• Data transformation functions
• TwinMind import utility
• Verification scripts
• Post-migration checklist

Platform Comparison

Error Handling

Resources

• TwinMind Import API
• Data Export Best Practices
• Platform Comparison

Congratulations!

You have completed the TwinMind skill pack. For questions, visit TwinMind Support.