AI must RTFM: Why technical writers are becoming context curators

I’ve been noticing a trend among developers that use AI: they are increasingly writing and structuring docs in context folders so that the AI powered tools they use can build solutions autonomously and with greater accuracy. They now strive to understand information architecture, semantic tagging, docs markup. All of a sudden they’ve discovered docs, so they write more than they code. Because AI must RTFM now.

It’s docs-driven development. It’s also technical writing. We should welcome our colleagues into the fold of technical communication and seriously start thinking about becoming context writers and maintainers. In a way, we’ve always been that, building the skills and techniques that allow owners of brains – either organic or simulated – to find their way in complex systems and accomplish tasks.

All AI requires to do the right thing is great context and a gentle nudge

Picture large language models (LLMs) as elaborate machines that take inputs (in most cases it’s just text), turn it into discrete pieces (tokens) and process it through an incredibly convoluted conveyor belt. At the other end of the machine is the output, usually in the form of helpful commentary, feedback, and commands issued to various system tools. The Chinese room, finally incarnated.

The quality of the output is a function of the quality of the input. We enter requests and get back what an average of all human thoughts could have led us to, eventually. There’s nothing magic to it: if you write clear, accurate, and well structured requests (prompts), chances are that the LLM will respond in kind. It’s hard work, but it pays off, because development time is substantially reduced.

That’s why, with every release of a new LLM, coders pay close attention to the size of the context window, that is, the amount of information one can feed to an LLM. A context window of one million tokens means you can easily feed the entire The Lord of the Rings trilogy to, say, Gemini, and start asking questions about it. And you would still have room for The Hobbit and The Silmarillion.

What is a Context curator and why we need that role

Engineers are finding out that writing, that long shunned soft skill, is now key to their efforts. In Claude Code: Best Practices for Agentic Coding, one of the key steps is creating a CLAUDE.md file that contains instructions and guidelines on how to develop the project, like which commands to run. But that’s only the beginning. Folks now suggest maintaining elaborate context folders.

A context curator, in this sense, is a technical writer who is able to orchestrate and execute a content strategy around both human and AI needs, or even focused on AI alone. Context is so much better than content (a much abused word that means little) because it’s tied to meaning. Context is situational, relevant, necessarily limited. AI needs context to shape its thoughts.

In Own the prompt, where I described how I built a tool to write docs using context from the terminal or the browser, I argued that technical writers should lead the way and own AI-powered docs processes, including the curation of context. In my predictions for this year (and probably the next, too), the importance of context was already present as the rise of docs-as-data.

Picture a developer inserting a docs cartridge into his AI powered code editor: the presentation layer is going to be largely irrelevant, the docs powering the answers of locally executed LLMs to aid developers in their coding quests. In a multi-channel content strategy, LLM-tailored output is going to be an additional, incredibly relevant channel.

Writing is designing and co-developing (again)

Four years ago, I argued that technical writers can play a key role in API design and development, because words are everywhere and we writers are uniquely well positioned to select the right words. Back then, OpenAPI was the device that allowed the magic of crafting design using just words. Today, the spell extends to all kinds of software development. We can conjure software ourselves now.

At this point, the most bleeding edge tech writing shops are serving llms.txt files and LLM-optimized Markdown. We’d take this a step further and prepackage context in a way that LLMs can easily consume. A standard for this still doesn’t exist, but I can see some flavor of DITA or another markup making a comeback, together with UIs that let users download which docs, for which version, etc.

Repomix allows anybody to select and package code and docs for LLMs

The endgame is to be able to make content accessible to LLMs and humans alike, to let them extract knowledge tailored to their needs. Tech writers become context writers when they put on the art gallery curator hat, eager to show visitors the way and help them understand what they’re seeing. It’s yet another hat, but that’s both the curse and the blessing of our craft: like bards in DnD, we’re the jacks of all trades that save the day (and the campaign).