The Entity Graph: Why Cross-Product Context Matters

Every SaaS tool creates its own version of your contacts. The same "Acme Corp" appears as a donor in your NGO tool, a client in your governance platform, and an employer in your HR system. Three separate records, three separate realities.

ORIS Entity Resolution uses deterministic matching (PAN, Aadhaar, CIN exact matches at confidence 1.00), probabilistic matching (Jaro-Winkler name similarity at 0.92+ threshold with confidence 0.80-0.99), and manual review for scores between 0.70-0.80 flagged for human verification.

The relationship graph connects entities through typed edges: director_of, donor_to, beneficiary_of, shareholder_in, employee_of, advisor_to, trustee_of, heir_of, spouse_of, and more. Each edge tracks the source vertical, effective dates, and a confidence score.

When ORIS AI receives a cross-vertical query, three gates must pass: (1) same tenant, (2) user has a role in the target vertical, (3) JWT includes the cross_vertical_access flag. This ensures intelligence flows only where authorized.