gamma-migration-deep-dive

What this skill does

# Gamma Migration Deep Dive

## Current State

!`npm list 2>/dev/null | head -10`

## Overview

Migrate presentation workflows from PowerPoint, Google Slides, Canva, or other platforms to Gamma's AI-powered generation. Gamma takes a fundamentally different approach -- instead of manually placing slides, you provide content and Gamma generates the presentation. Migration is about converting your content pipeline, not your slide files.

## Prerequisites

- Gamma API access (Pro+ plan)
- Source presentations accessible for content extraction
- Node.js 18+ for migration scripts
- Completed `gamma-install-auth` setup

## Migration Approaches

| Approach | When to Use | Effort |
|----------|-------------|--------|
| Content extraction + regeneration | Lots of text-heavy presentations | Medium |
| Import via Gamma UI | One-off migration of key decks | Low |
| Template recreation | Repeatable presentation formats | Medium |
| Parallel operation | Gradual transition over time | Low |

**Key insight:** You don't "import" slides into Gamma. You extract content from old presentations and regenerate them using Gamma's AI. This often produces better results than the originals.

## Instructions

### Step 1: Inventory Source Presentations

```typescript
// scripts/inventory-presentations.ts
import { readdir, stat } from "node:fs/promises";
import { join, extname } from "node:path";

interface PresentationInfo {
  path: string;
  format: string;
  sizeMB: number;
  lastModified: Date;
}

async function inventoryPresentations(dir: string): Promise<PresentationInfo[]> {
  const entries = await readdir(dir, { recursive: true });
  const presentations: PresentationInfo[] = [];

  for (const entry of entries) {
    const ext = extname(entry).toLowerCase();
    if ([".pptx", ".ppt", ".key", ".pdf", ".md"].includes(ext)) {
      const fullPath = join(dir, entry);
      const info = await stat(fullPath);
      presentations.push({
        path: fullPath,
        format: ext,
        sizeMB: info.size / (1024 * 1024),
        lastModified: info.mtime,
      });
    }
  }

  console.log(`Found ${presentations.length} presentations:`);
  const byFormat = presentations.reduce((acc, p) => {
    acc[p.format] = (acc[p.format] || 0) + 1;
    return acc;
  }, {} as Record<string, number>);
  console.log("By format:", byFormat);

  return presentations;
}
```

### Step 2: Extract Content from PowerPoint

```typescript
// scripts/extract-pptx.ts
// Use 'pptx-parser' or 'officegen' to extract text content

import JSZip from "jszip";
import { readFile } from "node:fs/promises";
import { DOMParser } from "xmldom";

async function extractPptxContent(pptxPath: string): Promise<string[]> {
  const buffer = await readFile(pptxPath);
  const zip = await JSZip.loadAsync(buffer);

  const slides: string[] = [];
  const slideFiles = Object.keys(zip.files)
    .filter((f) => f.match(/ppt\/slides\/slide\d+\.xml$/))
    .sort();

  for (const slideFile of slideFiles) {
    const xml = await zip.file(slideFile)!.async("string");
    const doc = new DOMParser().parseFromString(xml);
    // Extract all text elements
    const textNodes = doc.getElementsByTagName("a:t");
    const texts: string[] = [];
    for (let i = 0; i < textNodes.length; i++) {
      const text = textNodes[i].textContent?.trim();
      if (text) texts.push(text);
    }
    slides.push(texts.join("\n"));
  }

  return slides;
}

// Convert extracted content to Gamma prompt
function slidesToGammaPrompt(slides: string[], title: string): string {
  let prompt = `${title}\n\n`;
  slides.forEach((content, i) => {
    prompt += `Slide ${i + 1}:\n${content}\n\n`;
  });
  return prompt;
}
```

### Step 3: Batch Migration Script

```typescript
// scripts/migrate-to-gamma.ts
import { createGammaClient } from "../src/client";
import { pollUntilDone } from "../src/poll";
import pLimit from "p-limit";

const gamma = createGammaClient({ apiKey: process.env.GAMMA_API_KEY! });
const limit = pLimit(2); // Max 2 concurrent generations

interface MigrationItem {
  title: string;
  content: string;
  sourceFile: string;
}

async function migrateBatch(items: MigrationItem[]) {
  const results = await Promise.allSettled(
    items.map((item) =>
      limit(async () => {
        console.log(`Migrating: ${item.title}`);
        const { generationId } = await gamma.generate({
          content: item.content,
          outputFormat: "presentation",
          textMode: "condense", // AI condenses extracted text
          exportAs: "pptx",    // Get PPTX for comparison
        });

        const result = await pollUntilDone(gamma, generationId);
        return {
          title: item.title,
          sourceFile: item.sourceFile,
          gammaUrl: result.gammaUrl,
          exportUrl: result.exportUrl,
          creditsUsed: result.creditsUsed,
        };
      })
    )
  );

  // Report
  const succeeded = results.filter((r) => r.status === "fulfilled");
  const failed = results.filter((r) => r.status === "rejected");
  console.log(`\nMigration complete: ${succeeded.length} succeeded, ${failed.length} failed`);

  for (const r of results) {
    if (r.status === "fulfilled") {
      console.log(`  OK: ${r.value.title} → ${r.value.gammaUrl}`);
    } else {
      console.log(`  FAIL: ${r.reason}`);
    }
  }
}
```

### Step 4: Template Recreation

For recurring presentation types (weekly reports, proposals, etc.), create Gamma templates:

```text
Migration steps for templates:
1. Identify repeating presentation formats in your org
2. Create a one-page template gamma in the Gamma app:
   - gamma.app → Create → design a single representative page
3. Note the template gamma ID from the URL
4. Use POST /v1.0/generations/from-template with the gammaId
5. Update your automation scripts to use generateFromTemplate()
```

```typescript
// After template creation in Gamma UI
const MIGRATED_TEMPLATES: Record<string, string> = {
  "weekly-report": "gamma_template_weekly_abc123",
  "sales-proposal": "gamma_template_proposal_def456",
  "team-update": "gamma_template_update_ghi789",
};

async function generateFromMigratedTemplate(
  templateKey: string,
  content: string
) {
  const gammaId = MIGRATED_TEMPLATES[templateKey];
  if (!gammaId) throw new Error(`Unknown template: ${templateKey}`);

  const { generationId } = await gamma.generateFromTemplate({
    gammaId,
    prompt: content,
    exportAs: "pdf",
  });

  return pollUntilDone(gamma, generationId);
}
```

### Step 5: Validation Checklist

After migrating each presentation:

```text
- [ ] Content accuracy: AI-generated text matches source intent
- [ ] Slide count: reasonable for the content volume
- [ ] Theme/branding: workspace theme applied correctly
- [ ] Export quality: PDF/PPTX downloads successfully
- [ ] Links preserved: any URLs from original are in the content
- [ ] Stakeholder review: key presentations reviewed by owners
```

## Supported Migration Paths

| Source | Method | Fidelity | Notes |
|--------|--------|----------|-------|
| PowerPoint (.pptx) | Extract text → regenerate | Content-high, design-new | AI redesigns slides |
| Google Slides | Export as .pptx → extract | Content-high, design-new | Export first |
| Canva | Export as .pdf → extract text | Medium | Limited text extraction |
| Keynote (.key) | Export as .pptx → extract | Content-high, design-new | Export first |
| Markdown (.md) | Direct use as content | High | Best migration path |
| Notion pages | Export as .md → use directly | High | Clean text extraction |

## Error Handling

| Issue | Cause | Solution |
|-------|-------|----------|
| Content too long | Exceeds 100K token limit | Split into multiple presentations |
| Credit budget exceeded | Too many migrations at once | Batch over multiple days |
| Poor output quality | Content too unstructured | Add structure (headings, bullets) to extracted content |
| Missing images | Images in source not extracted | Gamma generates new images; reference image concepts in text |

## Resources