A philosophy professor at UC Berkeley built one of the most useful tools on the internet.
Not a developer. Not a startup founder. Not a tech company. A philosophy professor named John MacFarlane.
He wrote it in Haskell. In 2006. He still maintains it today. Twenty years later.
It is called Pandoc. The Swiss Army knife of documents.
You give it any document in any format. It converts it to any other format. One command. Seconds.
Word to PDF. Done.
Markdown to PowerPoint. Done.
LaTeX to Word. Done.
HTML to ePub. Done.
Jupyter notebook to PDF. Done.
Word to Markdown. Done.
PowerPoint to Markdown. Done.
Excel to Markdown. Done.
50 plus formats. Any direction. One tool.
pandoc thesis.docx -o thesis.pdf
That is it. One line. Your 300-page thesis converted. Formatting preserved.
Here is what this replaces:
A PhD student at Berkeley converting a thesis from LaTeX to Word. By hand: 6 hours. Pandoc: 3 seconds.
A novelist in Brooklyn converting a manuscript to ePub for Kindle. By hand: two days. Pandoc: one command.
A startup in Chicago migrating 200 HTML blog posts to Markdown. By hand: a full week. Pandoc: 4 minutes.
A professor at MIT turning 200 Markdown lecture notes into a PowerPoint deck. By hand: 2 hours. Pandoc: one command.
Here is what online converters charge for the same work:
Zamzar: $25 a month.
CloudConvert: $8 a month.
Smallpdf Pro: $15 a month.
Adobe Acrobat Pro: $239.88 a year.
Every one of them uploads your documents to their servers. Your thesis. Your manuscript. Your private notes. Sitting on someone else's machine.
Pandoc runs on your laptop. Nothing uploaded. Nothing sent anywhere.
44,997 stars on GitHub. GPL-2.0 license. Version 3.10 shipped June 4 2026.
Here is the wild part.
Pandoc powers R Markdown. Pandoc powers Jupyter Book. Pandoc powers Quarto. Every time you export a notebook to a PDF or render a research report, Pandoc is probably running underneath.
The standard of document conversion on the internet was built by one philosophy professor in his spare time.
In 2006, John wanted to write his lecture notes in a lightweight format and export them to Word, HTML, and PDF. He had never written a line of Haskell in his life. Pandoc was his first Haskell program.
He has shipped 14,505 commits since then. He co-wrote the CommonMark standard. He still teaches philosophy at Berkeley. He still ships Pandoc himself.
Microsoft did not build it. Adobe did not build it. Google did not build it.
One philosophy professor. Twenty years. Forty-five thousand stars.
Your documents. Your formats. Your machine. One command.
The week you used to lose to file conversion is back in your hands.
(Link in the comments)
THIS GUY BUILT 5 FAILED PRODUCTS.
Then he realized the problem wasn't the code.
It was the design.
Most founders think:
"If I build it, people will come."
They don't.
Because users don't just judge what a product does.
They judge how it feels.
So after 3 years and 5 failed attempts, he built something different.
An AI tool designed around a simple idea:
AI can generate.
But can it develop taste?
Instead of creating random screens, it tries to design an entire product with a consistent visual identity.
That's a much harder problem.
And if they solve it...
A lot of design tools are in trouble.
We're entering a world where building is cheap.
Taste is expensive.
Website:
https://t.co/Wjt3nAfGDe
Would you trust AI to design your next product?
El estadio Monumental de Universitario de Deportes es el recinto deportivo más grande y hermoso del país; es normal que la Conmebol la tenga en cuenta para organizar una final de Copa Libertadores, generando nuestra alegría y la envidia de otros.
Orgullosos de nuestro club.
Navego por Internet diariamente.
Aquí hay 10 sitios web increíblemente útiles que probablemente no sabías que existían:
Yo lo marcaría como favorito...