Causal Graphs (DAGs)
Each AICompass module uses an explicit causal structure. These directed acyclic graphs (DAGs) show which factors influence which outcomes — and where we apply intervention logic.
product fit
Strategic alignment, technical fit, and board readiness influence implementation success. We apply causal adjustment to vendor ROI claims based on your institution's confounding factors.
Edges (from → to)
- priority_match → strategic_alignment
- member_relevance → strategic_alignment
- timing → strategic_alignment
- board_readiness → strategic_alignment
- core_integration → technical_fit
- strategic_alignment → implementation_success
- technical_fit → implementation_success
- vendor_claimed_roi [confounding adjustment] → adjusted_roi
investment
Fit, ROI trajectory, and risk combine into a venture signal. We model how these factors affect investment success and exit potential.
Edges (from → to)
- fit_score → vc_demand
- roi_trajectory → vc_demand
- roi_trajectory → exit_potential
- risk_posture → round_risk
- risk_posture → vc_demand
- vc_demand → investment_outcome
- exit_potential → investment_outcome
- round_risk → investment_outcome
ma
Scale, cost synergy, and revenue synergy drive deal value. We model what happens to efficiency and ROE if we intervene on merge, headcount, or branch consolidation.
Edges (from → to)
- merge → headcount_reduction
- merge → branch_consolidation
- merge → scale_economies
- headcount_reduction → cost_synergy
- branch_consolidation → cost_synergy
- scale_economies → efficiency_ratio
- cost_synergy → efficiency_ratio
- revenue_synergy → roe
- efficiency_ratio → roe
- cost_synergy → npv
- revenue_synergy → npv
portfolio
Macro factors (rates, VIX) drive asset returns. CPA decomposes observed covariance into regime-invariant interventional component (Σ_do) and confounded component. Most valuable for regime robustness, not static optimization.
Edges (from → to)
- rates → asset_returns
- volatility → asset_returns
- inflation → asset_returns
- asset_returns → sigma_obs
- asset_returns [causal decomposition] → sigma_do
Honest framing
- CPA (Portfolio) — Genuine causal decomposition (Σ_do); most valuable for regime robustness.
- M&A — Causal attribution + do-calculus (E[Efficiency | do(merge)]).
- Product Fit — MCDA (multi-criteria decision analysis) + causal ROI adjustment for confounding.
- Investment (CVC) — Composite heuristic with explicit decision structure; not formal causal inference.