Document

Turn any PDF into a designed slide deck

Upload a PDF — a whitepaper, report, ebook, research paper, or pitch — and BananaSlides extracts the text, builds an outline, and renders a complete designed deck.

Free credits to try · No credit card required

What it does

PDFs are dense. Decks are skimmable.

PDFs are the dump format for everything serious — research, reports, contracts, sales material, internal docs. They're great for archival and reading; they're terrible for presenting. Asking a room to follow along while you scroll a PDF is a guaranteed lose.

PDF to BananaSlides solves that one-way conversion. You upload directly from your browser to Cloudflare R2, our server pulls the file, extracts the text with a Workers-compatible PDF library, and feeds it through the outline-and-render pipeline. The original file is deleted after extraction — we only need it long enough to read.

Common use: a research paper becomes a 15-slide explainer for the rest of the team, a whitepaper becomes a sales pitch, a policy PDF becomes a training deck. The extracted text quality is on par with copy-pasting from a PDF reader, only it's structured for slides instead of paragraphs.

How it works

Three steps from source to deck

Upload a PDF

Drag and drop or click to browse. Files go directly to Cloudflare R2 via a short-lived presigned URL — they never sit on our app servers.

Drop a PDF here, or click to browse

Max 50MB · stored on Cloudflare R2, deleted after extraction

whitepaper.pdf

1.4 MB

We extract text

A Workers-native PDF parser pulls the body text page by page. Scanned PDFs without OCR will produce limited results — text-based PDFs are ideal.

AI Outline

Draft

1Title slide
2Why this matters
3Key point #1
4Key point #2
5Supporting data
6Counter-arguments
7What to do next

Render the deck

Outline → slide-by-slide image generation. Pick theme, style, and aspect ratio up front, get a coherent deck out.

Designed deck

Features

What you get

Direct-to-R2 upload

Files upload straight to Cloudflare R2 via presigned URL. They never pass through your browser to our app servers.

Workers-native parsing

Built on unpdf, a pure-JS PDF parser that runs on Cloudflare Workers without needing Node's filesystem.

Up to 50MB per upload

Most reports, papers, and whitepapers comfortably fit. For larger files, split into chapters and run each separately.

Auto-cleanup

We delete the uploaded PDF from R2 as soon as the extraction succeeds. A lifecycle rule sweeps stragglers after 24 hours.

Preserves paragraph flow

Text is merged across pages with paragraph breaks intact, so the outline step doesn't fragment your content.

Saved to your account

Every generated deck shows up in your dashboard for re-download, sharing, or rebuilding.

Who uses it

Real use cases

Research paper explainers

Take a 20-page academic paper and turn it into a 12-slide internal explainer the rest of the team will actually read.

Whitepaper to sales deck

Marketing whitepapers are great for SEO, terrible for sales calls. Convert one into the deck your reps can actually present.

Compliance training

Internal policy PDFs become trainable slides for new hires, without anyone manually building the deck in PowerPoint.

Board meeting briefs

Annual reports and quarterly briefs travel as PDFs. Generate a deck version for the board meeting itself.

Manual vs BananaSlides

Why bother automating?

The manual version of this workflow is hours of copy, paste, restructure, design. Here's what changes:

Aspect	Manual	BananaSlides
Time from PDF to deck	2–4 hours per report	2 minutes for a 10-slide deck
Text extraction	Copy-paste page by page	Single upload, full body extracted
Structure	You re-outline by hand	AI weights ideas by importance
Design	Drop into a template	Themed image per slide
File handling	PDF sits in your downloads forever	Deleted from R2 after extraction

The reasoning

Why we extract on the server, not in your browser

Client-side PDF parsing exists, but it has real limits: a 50MB file can crash mobile browsers, encrypted PDFs need libraries the browser doesn't always have, and the parsing JavaScript adds megabytes of bundle weight to the page.

By uploading directly to Cloudflare R2 with a presigned URL and parsing on the Worker side, your browser only ever sends bytes — no heavy code runs locally. The Worker has the muscle to handle large files, and you don't pay a bundle-size tax for the privilege of opening a tool you might not need.

We delete the file from R2 right after extraction succeeds. Even if we miss, an R2 lifecycle rule sweeps anything older than 24 hours, so your source documents don't accumulate.

Pro tips

Get better decks from PDFs

PDF quality varies wildly. A few prep steps make a big difference.

Use text-based PDFs, not scans

If you scanned a paper document into a PDF, the text isn't selectable and we can't read it. Run OCR (e.g. via Adobe or a free tool) first.

Strip cover pages and TOCs

Long appendices and tables of contents dilute the outline. If your PDF has a lot of boilerplate, copy the meaty sections into a clean PDF first.

Match slide count to length

A 10-page report wants 8–12 slides; a 50-page whitepaper wants 15–20. Going much higher tends to produce repetition.

Pass a brand reference image

Adding a logo or product screenshot as a reference image keeps the visual identity tied to the original document.

FAQ

Common questions

Is my PDF secure?

Yes. Uploads go directly to Cloudflare R2 over HTTPS via a short-lived presigned URL. The file is private, scoped to your user, and deleted by us right after the extraction step. An R2 lifecycle rule sweeps anything we miss after 24 hours.

Does it work with scanned PDFs?

Only if they have an OCR text layer. Pure image-based scans contain no extractable text, so we won't have anything to outline. Run OCR with Adobe, Apple Preview, or a free tool before uploading.

What's the file size limit?

50MB per upload. Most whitepapers, reports, and ebooks comfortably fit. Larger files should be split — chapters tend to produce better, more focused decks anyway.

Can it handle password-protected PDFs?

No. Encrypted PDFs aren't readable by our parser. Remove the password (in Acrobat or Preview) before uploading.

Will tables and figures come through?

Text inside tables comes through as text. Images and figures don't — we generate fresh visuals on each slide. If a specific figure is critical, save it as a separate image and use it as a reference image when generating.

What's the credit cost?

One credit per slide generated. Extraction itself is free. So a 10-slide deck from a 50-page PDF costs 10 credits, same as a 10-slide deck from a one-line prompt.

Do I need an account?

Yes. Generating slides consumes credits, and credits are tied to your BananaSlides account. New accounts get a small free pack so you can try the tool end-to-end without paying. You can sign up with email or Google in under thirty seconds, and the source you've already entered (URL, file, or prompt) is preserved across the redirect.

Can I edit the slides after generation?

Each generated slide is delivered as a high-resolution image. You can download individual slides, or download the whole deck as a ZIP and drop the images into PowerPoint, Keynote, Google Slides, or Notion as a starting point. For text edits, regenerate the slide with a tweaked outline — the deck is fast enough that iterating beats hand-fixing pixels.

What is the BananaSlides format?

BananaSlides decks are designed to be visual, not bullet-heavy. Each slide is rendered end-to-end by an image model so you get a real layout — typography, illustration, color — instead of a generic template with your text dropped in. You choose the theme, style, and aspect ratio up front, and the AI keeps every slide visually consistent across the deck.

Related tools

Ready to try PDF to BananaSlides?

Scroll back up, drop in your source, and ship a deck in two minutes.

Start generating Browse all tools