AI-Assisted Study Design
AI drafts and organizes; experts decide. Every AI contribution on Opendelphi is grounded, cited, and gated behind human consensus.
Study Architect
Describe your study goals in plain language and the Study Architect agent drafts candidate instruments — retrieving from a corpus of validated research instruments and proposing items grounded in what it found. It's a conversation, not a black box: you steer, question, and revise.
Citation Binding, Validated
Corpus-grounded items cite the specific corpus entries that ground them, and those citations are validated against the actual retrieval set — failing citations are rejected. When the corpus has no coverage, drafts can ground in current literature via web search or proceed ungrounded, clearly flagged for moderator review.
Clustering, Not Anchoring
In open Delphi rounds, experts generate items first and AI clusters the submissions into themes for the panel to review. AI never pre-seeds the panel's thinking by default — that ordering is a deliberate methodological guard.
Humans Hold the Gate
No AI output becomes a published instrument on its own. Drafts go through the Delphi process, and only panel consensus materializes an instrument for deployment. Licensed-scale text is filtered at generation time, and AI features operate on study design content — not on participant response data.