Our principles

We have an initial set of principles guiding our work in this community. These may evolve over time.

• Human agency and accountability

  Agents should support researchers, not replace judgement, responsibility, intellectual ownership and accountability.

• Value and meaning over AI Slop

  The goal is not simply to produce more text, code, experiments, or papers. The automation should help us in creating more value and meaningful work.

• Safety and reliability

  Research agents should be reliable, transparent about their sources and uncertainty, and designed so that their outputs can be checked and traced.

• Accessible and open

  Useful agentic tools should be accessible to institutions and researchers even with limited resources or funding. We encourage open-source tools, open-weight and low-cost models, and solutions that can adapt to different disciplines.

• Community-driven

  The future of agentic academia should be shaped by researchers themselves. We value shared tools, open collaboration, diverse perspectives, and solutions developed around real academic needs.