Getting Started: Edit Your First PDF in 5 Minutes

A hands-on walkthrough of the complete editing workflow — from setup to download.

This tutorial walks you through your first complete edit using Notebook LM Slide Editor: getting a free Gemini API key, uploading a PDF, replacing text with AI-matched style, and downloading the finished file. Total time: about 5 minutes.

Before You Begin: Get a Free Gemini API Key

Notebook LM Slide Editor uses Google Gemini's multimodal AI to analyze text typography in your documents. You need a free Gemini API key before running your first OCR analysis. Here's how to get one in under 2 minutes:

Go to aistudio.google.com and sign in with your Google account.
Click "Get API key" in the left sidebar, then "Create API key."
Choose "Create API key in new project" if prompted.
Copy the key that appears — it starts with AIza and is about 39 characters long.
Return to Notebook LM Slide Editor and click the Settings gear icon in the top navigation bar.
Paste your API key into the "Gemini API Key" field and click "Save."

The Gemini API free tier provides more than enough quota for personal and small-team use. No credit card or payment method is required to obtain or use the free tier key.

Step 1: Upload Your Document

Notebook LM Slide Editor accepts PDF, PNG, JPG, and WebP files. To upload your first document:

Drag and drop your file directly onto the large upload area in the center of the main page.
Or click "Select File" to open the system file picker and browse to your file.

Multi-page PDFs are automatically split into individual slide images and displayed in the thumbnail sidebar on the left. Click any thumbnail to jump to that page. Single images (PNG, JPG, WebP) appear as a single page.

Resolution tip: For best OCR results, use source files at 300 DPI or higher. When exporting from Google Slides or PowerPoint, choose the "High Quality" or "Print Quality" PDF option. Higher resolution gives the AI more pixel detail, which directly improves font and color detection accuracy.

Step 2: Select the Text You Want to Replace

Once your document is open in the editor, navigate to the page you want to edit using the thumbnail sidebar. Then:

Click and drag your mouse over the text area you want to replace. A blue selection rectangle appears as you drag.
Make the selection box slightly larger than the text itself — include some of the surrounding background. The AI uses the background color to match the replacement text to the original.
Release the mouse. The selection stays active and the right sidebar activates with style controls.

If you mis-select an area, just click and drag again anywhere on the canvas to create a new selection. The previous selection is replaced automatically.

Step 3: Run AI OCR Analysis

With your selection active, click the "Run AI OCR" button in the right sidebar (the button with the wand or sparkle icon). The editor sends the selected image region to the Google Gemini API for analysis. After 2–5 seconds, the sidebar populates with detected values:

Font Family — the closest matching font family (e.g. Inter, Roboto, Helvetica Neue).
Font Size — the detected size in pixels.
Font Weight — Regular, Medium, SemiBold, Bold, etc.
Text Color — the exact hex code of the original text color.
Background Color — the color behind the text in the selected region.
Letter Spacing — the detected tracking in pixels.
Alignment — Left, Center, Right, or Justify.

Review these values. The AI is accurate for most standard fonts at normal reading sizes. If a value looks incorrect — for example, the weight shows "Regular" but the original text is clearly bold — adjust it manually using the dropdown or slider before proceeding.

Step 4: Type Your Replacement Text

In the Content field in the sidebar, type or paste your replacement text. A live preview overlay appears on the canvas showing the new text rendered in the AI-detected style. As you type, you can fine-tune:

Font size — increase if your new text is shorter than the original; decrease if it's longer.
Letter spacing — the spacing slider is especially important for headings. Even a 1–2px difference is visible at display sizes. Compare your overlay against unedited text nearby to calibrate.
Alignment — use the Left / Center / Right buttons to match the original alignment.
Background transparency — if the text sits on an image or complex pattern rather than a solid color, toggle the background to transparent for a cleaner overlay.

When the overlay looks right, click "Apply Text" to commit the edit to the canvas. The overlay is now a permanent layer on the page, and you can continue editing other areas or navigate to other pages.

Step 5: Download Your Edited File

When you have finished all your edits across all pages, export the result:

Click "Download PDF" in the top navigation to export the complete document as a multi-page PDF. All overlaid text is baked into the output pages.
Click "Save Images" to export each page as an individual PNG file — useful when you want to insert specific slides into another application.

The file is generated entirely in your browser and saved to your default downloads folder. Nothing is uploaded to our servers. The process typically takes a few seconds for standard-length documents.

Troubleshooting Common First-Time Issues

The OCR button doesn't respond or returns an error

Check that your Gemini API key is correctly entered in Settings (gear icon). The most common issues are a missing key, an extra space, or a key that hasn't been activated yet. If the key is correct, wait 30 seconds and try again — Google's API occasionally experiences brief rate-limit bursts on new keys during their warm-up period.

The replacement text looks slightly different from the original

The most likely cause is a font mismatch. If the original document uses a custom or proprietary font not available in the browser, the AI substitutes the closest available alternative, which may have slightly different proportions. Try adjacent fonts in the font family dropdown to find the closest visual match. Also double-check the letter spacing — a 1–2px difference in tracking is the second most common cause of visible mismatch.

The PDF upload seems to hang

Very large PDFs (more than 150 pages) may take 20–40 seconds to render, especially on older devices, since all rendering happens locally in the browser. If it seems stuck, wait one full minute before concluding there's a problem. If the upload fails repeatedly, try splitting the PDF into smaller sections (50 pages or fewer) using any free PDF-splitting tool and uploading them separately.