Want to know where tech writing is headed? Watch Anthropic.

But before we dive into the job descriptions, let’s talk about the elephant in the room. Why does Anthropic need human writers at all? Shouldn’t they just generate their tech docs using AI?

There are many potential answers to that question, but the most obvious one in my mind is that AI cannot generate something out of nothing. As a company creating a truly new product, Anthropic needs someone to document the first iteration of the product content.

Having said that, I imagine that Anthropic can generate plenty of tech docs directly from their code. If their code is well-structured and their engineers include good inline comments (as I would expect from Anthropic), it should be relatively simple to use AI to generate content that defines the calls, parameters, variables, and defaults for APIs. Other content, such as setup procedures, could probably be at least partially generated with AI by providing it commands and formatting guidelines.

But a lot of the content is less straightforward. Things like best practices and customer use cases are locked up in human brains. This is high value content, and to make it available to customers, Anthropic will need human writers.

Plenty of companies use a combination of people and AI to create tech docs these days. So, what makes the Anthropic job description so different?

I should clarify that I’m assuming that Anthropic uses a combination of AI and humans to create their content, even though their job descriptions don’t say as much. It only makes sense that Anthropic writers would use AI though — both because Anthropic is a top-tier AI company and because they expect their writers to have a vast level of expertise.

For instance, if you take a closer look at the job in the Claude Code team, you’ll notice that they’re looking for someone with deep expertise in writing, content strategy, and engineering. The content they’ll expect the writer to create requires both strong product expertise and in-depth knowledge of their customer base. This depth and breadth of knowledge usually represents the skills of multiple people. Anthropic is looking for all of it in one person. It’s common these days to ask for the moon in job postings, but this goes farther than anything else I’ve seen in the tech writing field.

Most of us are already using AI to create at least some content. We can expect businesses will to lean into that trend, particularly as small language models make AI more efficient. Using AI for content that can be majority-derived from product code and structured templates will be a no-brainer for many organizations. If it’s cheaper and faster than humans, why wouldn’t they use it? We can also expect businesses to use AI to regression test content, update existing content, and fix bugs.

That leaves humans confirming that the AI-generated docs are correct (hopefully) and creating the docs that are too difficult for the AI (such as best practices for using the product, the quirks and corner cases that AI cannot predict, and use cases based on customer conversations).

Many companies will be tempted to assign these tasks to engineers, and have AI edit the results. But technical writing is a unique (although often undervalued) skillset. As is true today, authors will need to be able to put themselves in the users’ shoes, intuit what they need to know, and then communicate it in a concise and clear manner. AI can help people do this, but cannot fully replace the empathy and experience required to do it well.

If you’re a tech writer who is concerned that you don’t have the required skills for this bold new future, there’s plenty you can do to prepare.

Know the value of your existing skills. Writing skills are often undervalued, leaving you at a disadvantage in the job market. When done well, writing brings consistent value to an organization. Learn how to speak to people outside the writing profession about the efficiencies and competitive advantages that strong writing brings to your business.

Grow your skills. Assuming I’m right, this new future will really put the “technical” back into technical writing. If you haven’t already, at least learn the basics of a programming language. Consider learning about SQL or noSQL databases. Ideally, you would earn certificates as you go to show your progress, but the most important thing to do is keep learning.

Experiment with AI. Use the tools at your disposal and see what they can do. AI is a great tool for helping you get work done faster, but you can also use it to help you as you grow your skills. (Personally, if I’m having trouble grasping a new concept, I’ll ask AI to reiterate it a few other ways to see if one of them resonates with me.) If you feel like your AI learning has stalled, consider setting up a weekly brainstorming session with your peers where you experiment with using AI to solve your day-to-day issues.

Stay curious. Our industry is changing daily. Your biggest asset is curiosity. Follow new developments. Think how they apply to you. Talk about them with your peers and keep experimenting.