Specs and Docs ⋅ Markdown and Large Language Models: a perfect combination

LOOPS, our publishing workflow for product literature, employs both traditional and bleeding-edge technology. At both ends of this spectrum, you’ll find extremes that interact surprisingly well:

The rise of LLMs

Unless you have been living under a rock for the last two years, you know that generative artificial intelligence is disrupting many areas of human creativity and productivity, from education and research, journalism and technical writing to literature. People use large language models to research complex topics, create and improve documents, write poems, and engage in conversations about topics both banal and profound. Most people will use the respective LLM’s web interface, which usually provides well-structured answers with headings and lists. Important information is highlighted in bold type or italics.

“Waiter; there’s Markup in my Content Soup”

When users want to reuse content from Claude or ChatGPT elsewhere, they will usually click / tap the “Copy” icon. When pasting that content from the clipboard, they’ll see markers around words – *asterisks* and __underscores__, and they may wonder what these are / represent.

Well, these markers are Markdown; the lightweight text markup behind millions of documents that were created and published well before the advent of LLMs.

Users who have never worked with markup languages may gnash their teeth and strip away these markers before they spend time in Microsoft Word or another application to format the text before printing or exporting. Which is really a shame, because the formatting is already there. Markdown provides markup that is readable by humans and easy to process for many text editors, conversion tools, and content management systems.

For technical writers who use large language models to create or refine content, these markers (actually, “#”, “_”, hyphens, and brackets are sufficient for almost every use case) are a blessing, because they can speed up the publishing process significantly.

In a Markdown-based publishing workflow such as LOOPS, Markdown-formatted content copied from the web interface of a Large Language Model can be dropped directly into a text editor, resulting in a formatted and well-structured document with headings and lists.

There and back again

Even better: Most large language models can be prompted to accept and generate Markup code on the fly. This means that authors who want to polish, extend, or summarize their text, add links or other information to their text, can prompt the LLM accordingly and receive ready-to-use, well-formatted content.

Being able to work with formatted text across system borders (large language models, text editors, content management systems) is a huge timesaver.

Giving Search Engines Something to Chew On

Some large language models, such as Claude and ChatGPT, and new search engines such as Perplexity, can access web content (either directly or using previously cached results) and return it in digested form to the user. When you make your product literature available as well-formed and semantic HTML, these engines can process it more effectively, increasing product understanding and accordingly user satisfaction without investing a single cent.

This isn’t the stuff of Science Fiction. To see it in action, provide ChatGPT with the URL of a web manual and ask it to return a digest. After a few seconds, you should see a summary of the product’s features and basic operation instructions. You should also be able to extract information about a specific product feature or ways to troubleshoot a problem.

LLMs / Apple Intelligence

Using Apple Intelligence, macOS and iOS devices (i.e., Macs, iPhones, and iPads) can create summaries of well-formatted web manuals in Safari, giving users a quick overview of features, setup, and troubleshooting procedures.

In productivity apps such as Pages, Apple Intelligence can be used to summarize or restructure content. More features are expected for upcoming macOS and iOS releases.

A growing ecosystem

The ecosystem of applications that support both Markdown and large language models is expanding all the time. Notion (the “application for everything”) is another popular productivity and project management tool with AI support, allowing users to create and process formatted content using artificial intelligence, and Notion has supported Markdown from day one.

By the time you read this document, there will likely be more tools supporting Markdown and LLMs. With LLM integration into modern operating systems, you can expect even more efficient workflows. Let’s go exploring!

Ready to leverage Large Language Models for your manuals?

Let’s discuss how LOOPS can help you create better documentation.

↻ 2025-08-21