Docling: a notable open-source project worth a look

A field note on Docling, a notable open-source project we came across. What it does, where it might fit, and why we're watching it rather than claiming we use it.

Last reviewed July 25, 2026 new-and-notableopen-sourceai-basics

We keep an eye on new open-source projects in the areas we work in, and Docling came up recently. We have not put it into production, so treat this as a field note rather than a recommendation. Here is what it is and why it caught our attention.

What it is

A document conversion library that turns PDFs and office files into clean structured markdown, aimed at feeding documents into AI retrieval systems.

Where it might fit

For a small business the interesting question is never the tool itself. It is the job it does. Something like this tends to matter when you are trying to get messy source material into a shape an AI system can actually use, which is usually the unglamorous half of any real project. We would want to test it against real documents before trusting it with anything that counts.

Why we are watching, not adopting

A project looking promising on release day is a weak reason to build on it. We wait to see whether it stays maintained, whether it handles the awkward real cases, and whether it earns a place next to what we already run. Until we have run it ourselves, “notable” is the honest label rather than “we use this”.

Where to look next

If you want the groundwork this kind of tool plugs into, Building a knowledge base your AI can actually use covers getting your source material in order first. Source: https://github.com/DS4SD/docling