Gap Analysis · Gap Search Agent
Identifies defensible research gaps, contradictions, and candidate research questions from a user-curated collection of academic papers.
Designers: Nidhi Jha, Emily Foshee, Ankur Kumar, Venkatesh Kolluru
An evidence-grounded, non-authoritative agent with a human-in-the-loop workflow that acts as a neutral evidence synthesizer, providing structured output without independently assessing novelty, resolving contradictions, or judging scientific merit.
The Gap Agent is an AKD-designed agent that identifies defensible research gaps, contradictions, and candidate research questions from a user-curated collection of academic papers. This agent is evidence-grounded, non-authoritative, and incorporates a human-in-the-loop workflow. The agent acts as a neutral evidence synthesizer, providing structured output without independently assessing novelty, resolving contradictions, or judging scientific merit. Although the Gap Agent is also Stage 1 of the Closed-Loop CM1 pipeline, it can be used as a standalone tool.
The agent executes a six-stage pipeline with explicit human-approval gates between the stages. These stages include scientific scope inference, structured extraction with paragraph-level traceability, gap-matrix proposal across analytical lenses, gap identification (explicit, inferred, and contradictions), optional research question suggestions, and qualitative tiered prioritization (high/medium/exploratory). Users approve the output of each stage before the agent proceeds.