IdeaForge AI: A Dual-Mode, Auditable Research-Question Laboratory for Converting Scientific Ideas into Startup Opportunities

Universities face a paradox of abundance: artificial intelligence can generate research questions faster than institutions can investigate, validate, incubate, or commercialize them. This case study presents IdeaForge AI, a dual-mode, auditable research-question laboratory that converts this abundance into a decision process without claiming to predict scientific truth or startup success. The system operationalizes Discovery Plane Theory through Discovery Complexity and Question Compression, adds coordinate-equivalence handling and sensitivity analysis, and separates those research constructs from a commercialization screen. Manual Expert Mode uses anchored guidance and declared admissible procedures, while AI-Assisted Mode proposes editable values, representations, confidence, and rationales. Users may create portfolio sizes from one to ten questions. Reproducible tests recover the worked university-energy case, including q3 = (0.750, 0.800), Region IV, frontier membership, U0.5 = 0.425, and a 79/100 venture-screen value. A 196-setting sensitivity audit on the four-question worked portfolio retains q3 as highest-ranked throughout. Startup Studio then structures problem definition, users, paying customers, MVP design, pilot metrics, risks, business hypotheses, and next actions. IdeaForge therefore provides a human-gated bridge from question formation to evidence-seeking commercialization while preserving provenance, editability, and methodological boundaries. Its web interface exports portfolio CSV, JSON, and per-question PDF evidence for institutional review.