AI for Human-Mediated Communication
How large language models reshape the way people write, persuade, and connect when the model sits between them, and how to keep the human voice intact.
The Lab · SZD Labs Research
Open research and field notes from SZD Labs: AI for human-mediated communication, evaluation and reliability, and the agentic systems reshaping how we work.
We don't build AI to replace thinking. We build, study, and publish systems that extend it.
The threads we're studying and publishing as the work develops. New pieces land here.
How large language models reshape the way people write, persuade, and connect when the model sits between them, and how to keep the human voice intact.
Frameworks and evals that make agentic systems trustworthy enough to put in front of real users, not just demos.
What we learn shipping autonomous and semi-autonomous AI inside real organizations, from first pilot to daily driver.
New research threads open as the work develops. Have a hard problem worth studying? We take on applied research with partners.
Collaborate on applied research, or get in touch when a problem's worth studying.
Get in touch →