Writing for Different Readers: When Your Audience Is Half Human, Half LLM
Nobody actually reads documentation. Humans skim it in predictable, well-studied eye-tracking patterns — and AI assistants don't read at all; they chop content into chunks and reassemble answers. So who are we really writing for? This talk looks at how human eyes move across a page and how large language models ingest that same page — and lands on a surprising payoff: the fixes are mostly the same. Descriptive headings, self-contained sections, and front-loaded meaning create both the "layer-cake" scan humans love and the clean chunks machines need. Good information architecture isn't two competing jobs. It's one practice that serves bothreaders at once. Grounded in real product/developer documentation, real traffic data, and concrete before/after changes from daily practice.
Anastasiya Biaroza is a Senior Technical Writer at Box, where she creates both product and developer documentation with a focus on clarity, usability, and user experience. She enjoys turning complex technical concepts into content that helps people get the most out of a product. Her work spans AI-powered features and developer tools, and she is passionate about making technology more accessible. Outside of work, Anastasiya enjoys traveling, playing tennis, and learning languages. A fun fact: she is currently on a 907-day Duolingo streak while learning German.