AI May Free Us From the Tyranny of the PDF

Could AI spell RIP for the PDF?

Artificial Intelligence may bring about the end of the tyranny of the pdf. We want to help.

The cracks are already beginning to show. The army of computer scientists & data experts building LLMs and other AI tools are struggling to get the models to read old pdfs accurately. Government data in particular does not fare well through OCR (Ocular Character Recognition) - the technique used to extract the data from pdfs and make it machine-readable.

"Part of the problem is that PDFs are a creature of a time when print layout was a big influence on publishing software, and PDFs are more of a 'print' product than a digital one," Derek Willis, a lecturer in Data and Computational Journalism at the University of Maryland, wrote in an email to Ars Technica. "The main issue is that many PDFs are simply pictures of information."
‍

The solution to us seems simple: build systems that obviate the need for pdfs. Move us away from strategy and brand guidelines and science stored & shared in static pdf files. That's what we're doing with NOAN.

There's one problem: people f**king LOVE pdfs.

We've discussed pdfs at length while building NOAN. According to our Slack, we've shared 153 of them, which, frankly seems low. No matter what you're doing, there's no avoiding the lowly, super-functional pdf file. There are 2.5 trillion of them in existence. The format has been around for 30 years and it's going nowhere – because the pdf does, admittedly, do a basic job really well. It makes a document available to read online or print easily. It's universally shareable. It works on every platform. For what it does – an extremely lo-fi way of sharing a document – it works.

BUT – and here's where the friction occurs – each pdf is a walled garden. We've built technologies to get around that (OCR, which we're trying to replicate with AI) but workarounds suck. The content within a pdf is static and hard to edit and often messy, old, or formatted weirdly. When you do succesfully version a pdf, you end up with updated file name structures to signal that change to readers, and that's no way to live.

PDFs are great static documents. Maybe the best. They are terrible for dynamic collaboration, hence the emergence of Google Docs/Sheets or the seventh circle of hell that is Microsoft Sharepoint. Regardless, we're still using PDFs to share strategy because they are universal and relatively easy to distribute and also offer blunt control: the minions can't really edit them, they just have to read, interpret and apply them to their digital existence.

We're using a technology created before the DVD was invented to power our digital lives in 2025, folks.

If strategy was a static thing, a static doc would be fine, because strategy would never change.

But strategy demands change, and constant revision. Which is why you end up with users asking genuine questions like: Are we working off strategy_v2.pdf or strategy_v3.pdf? Or is it new_strategy_v2.5_USETHISONE_final.pdf? Madness.

There are two major flaws with the pdf as used for sharing dynamic strategy:

1) Any time you tweak your strategy, you end up recirculating and revising the newly amended 'live' strategy file, hoping others will successfully disregard the old versions and not accidentally refer to them in future. This is costly and stressful and ineffective.
2) Turning the pdf-bound strategy into action relies on human interpretation of the pdf. This is subjective and inconsistent – the more so the bigger you get.

This is why we built NOAN to create a structure for strategy – so that content creation is always drawing on the canonical, live form of your strategy. Then it leverages AI to create all the content directly from that strategy – in seconds – meaning the interpretation is consistent and can be done by anyone. Need to amend your strategy, change a tagline, add or remove products? No problem, make the edits in Build Mode, and everything you or your team create will be instantly aligned to the new changes.

As the poll shows, we love pdfs. We love to print 'em off, dip our toes into a soft pair of slippers, light a pipe and sip some camomile tea as we read them by candlelight, making notes in the margin with an inky quill. But we hate that they're still the dominant form of top-down strategy deployment in 2025.