One-Liner
An argument-structure review tool that flags logical weaknesses and anonymization leaks in Horizon Europe first-stage proposals, adapted for the new anonymous evaluation format.
AI Thinking Process
41 Horizon calls use two-stage submission; anonymised evaluations piloted; argument itself must carry proposal — not PI reputation.
PI prepares Horizon first-stage abstract, circulates to friendly colleagues, receives gentle feedback. No structural argument stress-test for anonymised evaluation context.
Verb Transplant: 'code review' from software → scientific proposal argument structure. Flags structural weaknesses (unstated assumptions, weak causal links, non-anonymised language leaks).
Three compounding risks: anonymised eval pilot could be rolled back, academic buying cycles slow, adoption barrier mid-strength. Conviction 43%.
Pivot to generic NIH/NSF/UKRI/ERC structural argument review tool sold institutionally. Adoption barrier persists. Platform extenders (Grammarly, Paperpal) dominant. Conviction 38%.
Below 40% threshold after pivot. Three compounding risks confirmed as durable ceiling.
Kill Reason
43% conviction below the 50% threshold — three compounding risks: (1) anonymised evaluation is a pilot that could be rolled back after the 2026–27 cycle, (2) academic buying cycles are slow, (3) adoption barrier is moderate (PIs may resist feedback that challenges authorship authority).
Risk Analysis
Risk analysis available for latest engine ideas.
What do you think?
Related ideas you can explore free:
killed: Classic form-filling automation: redaction is a narrow feature that research office platforms (Pure, Elsevier, Worktribe) can add in a sprint, and frequency is too low (20–50 proposals per year per office) to support a subscription product.
killed: 43% conviction below the 50% threshold — three compounding risks: (1) anonymised evaluation is a pilot that could be rolled back after the 2026–27 cycle, (2) academic buying cycles are slow, (3) structural adoption barrier is moderate (PIs may resist feedback that challenges authorship authority).
killed: Classic form-filling automation: redaction is a narrow feature that research office platforms (Pure, Elsevier, Worktribe) can add in a sprint, and frequency is too low (20–50 proposals per year per office) to support a subscription product.