Project
AI Investment Analysis System
Multi-stage AI system that replicates institutional investment workflows, from opportunity sourcing to structured portfolio review and critique.
Designed around role-based reasoning (analyst → senior → managing director) with structured evaluation and decision gating at each stage.
Built and tested as a working system where outputs from each stage feed into subsequent evaluation and decision steps.
System Behavior
- Generates broad opportunity sets at the initial analyst layer
- Filters and refines outputs through structured evaluation and consistency checks
- Applies adversarial critique and portfolio-level review to stress-test outputs and surface weaknesses
Architecture
Claude-based multi-stage pipeline with role-based evaluation, structured decision flow, and output handoff between stages.