Multi-Perspective AI Research Agent System

A qualitative research group (managed by Sabrina Lin) needed a way to analyze complex qualitative samples (TCS observations) without the inherent bias of a single human or AI perspective.

Qualitative research requires looking at the same data through multiple lenses—sensory, structural, cultural, and poetic. Manually coordinating these viewpoints across large datasets was slow and prone to 'tunnel vision'.

• Subjective Bias: Research analysis limited to one viewpoint
• High Complexity: Coordinating four distinct philosophical frameworks
• Scaling Friction: Analyzing thousands of field observations manually
• Data Inconsistency: Difficulty in quantifying 'tension' or variance between observations

Deep Loom developed a 'Parallel Curator' architecture using n8n and Claude 4.5. This system processes a single research input through four specialized AI agents simultaneously.

The workflow reconciles these diverse perspectives into a unified 'APR' (Average Perspective Rating) and automatically flags data points with high variance for human review.

• The Phenomenologist (B1): Analyzes lived experience and embodied sensation
• The Structural Objectivist (B2): Focuses on geometry, spatial logic, and verifiable structure
• The Cultural Contextualist (B3): Decodes social symbols and cultural performance
• The Poetic Synthesist (B4): Identifies metaphor, resonance, and thematic ambiguity
• Automated Synthesis: A centralized processing hub that calculates APR Tension and archives results to Notion

The system provides researchers with a multi-dimensional analysis that was previously impossible to achieve at scale. By quantifying 'tension' between curators, researchers can immediately identify which samples are most complex or ambiguous.

The automation has completely replaced manual data entry into Notion, ensuring a pristine audit trail for every observation and allowing the research team to focus on synthesis rather than categorization.