Carouselr

Why we built it

On LinkedIn, code is often posted as a screenshot. A screenshot is a picture: you cannot select the code, and small text is hard to read on a phone. LinkedIn’s document posts, the swipeable carousels, accept a PDF, and a PDF can hold real text.

Carouselr is built around that: each slide is designed for code first, and the PDF export keeps the code as text.

From one line to a carousel

You give a topic in one line. A language model from OpenAI drafts the slides: a title, the points, the code samples. Carouselr’s privacy page names that provider. The draft is a starting point: you edit every slide, its text and its code, before anything is exported.

Code blocks are highlighted in 30 programming languages, from Python and TypeScript to Rust, SQL and Bash. Slides come in portrait 4:5 at 1080 × 1350 px, plus square and 16:9.

A PDF that keeps the code as text

The PDF is generated on our server, not captured from the screen. Its text stays text: it can be selected and extracted, and the fonts are embedded in the file. PNG and HTML exports come with the Pro plan.

How it runs

The website runs on a Cloudflare Worker. The app, its API and the PDF service are deployed with Coolify, the data lives in a Supabase Postgres database, and images are stored on Cloudflare R2. Carouselr opened to the public in September 2026.

Polar sells the Pro plan on our behalf as the merchant of record: it takes the payment and handles the VAT for your country.

How it is tested

Every pull request runs the type checks, the linter, the API tests on an in-memory Postgres, the app and website tests, and a test render of a real PDF by Gotenberg, the same PDF engine production runs.

Visit carouselr.com