Notes from the team
Research updates, methods, and reflections from the otto-SR team.
We asked the frontier models behind ChatGPT, Codex and Claude Code to run systematic reviews end to end, and scored each stage. They can now do the whole review, but they aren’t ready to do it unsupervised.
We used the public Tarlatamab assessment to test how quickly and accurately otto-SR could reproduce the evidence behind seven unique PICOs.
Why study grouping breaks meta-analyses—and how otto-SR fixes it.
We’re launching an AI-powered platform that takes systematic reviews from months to hours.
