Technology
The Brazilian consórcio system processes 13 million members on infrastructure built decades ago. We are bringing the same model to the U.S. with the technology stack the original system never had — patent-pending AI underwriting, open-by-default APIs, and an audit trail that meets U.S. consumer-finance regulatory requirements.
Six layers, each designed around a specific guarantee we owe members: fairness, transparency, fund safety, and operational integrity.
A proprietary scoring model that ranks members for voucher selection based on contribution behavior, savings discipline, and engagement signals — never on credit score, race, gender, age, national origin, or any ECOA-protected attribute. The engine is trained, audited, and continuously monitored for fairness.
Every contribution, voucher, and obligation is recorded in an append-only ledger reconciled daily against the trust account at US Bank. The ledger is the system of record for member positions; Jackson Hole Trust Company independently verifies balances.
Cross-cluster behavioral analysis identifies coordinated inauthentic activity, contribution-pattern anomalies, and account-takeover signals — all aggregate, never individual-credit-decisioning. The fraud layer protects member contributions without touching protected-class attributes.
OpenAPI 3.1 spec, JSON-LD knowledge dataset, and Model Context Protocol (MCP) endpoints are all publicly available. AI agents and crawlers can read canonical content as data without scraping the rendered HTML.
A conversational agent that explains the model, runs personalized math, and walks members through enrollment in plain English (and Portuguese, Spanish). Built on Vertex AI with retrieval-grounded answers — never makes up fees, rates, or eligibility.
Every member can audit their own contribution history, voucher position, and Savings Score factors at any time. Every fee is disclosed before enrollment. The platform is built on the principle that transparency is the cheapest fraud-prevention tool any cooperative can deploy.
Defensible IP
Covering the savings-score AI architecture and behavioral-underwriting methodology. Granted by the USPTO. Defensible IP around the proprietary engine that ranks members for voucher selection.
Additional methodology for cooperative-financing risk modeling. In examination at the USPTO. Patent number on grant.
The original Brazilian consórcio system was built before AI, before behavioral analytics, and before regulatory frameworks like ECOA, GLBA, and FCRA existed. We are bringing the model to the U.S. with the technology and compliance scaffolding it always deserved.
Fairness commitment
The Savings Score uses an explicit feature allowlist — every input the model sees is documented, justified, and reviewed for disparate-impact risk before it lands in production. The model never sees credit score, race, gender, age, national origin, or any other ECOA-protected attribute. Period.
We monitor the model continuously for fairness drift across geography, asset class, and contribution behavior. The audit trail is end-to-end; every feature reaching the scoring engine is reviewable, and the methodology is reviewed by external counsel as part of our compliance program.
See exactly which factors the score usesRead the white papers, browse the public APIs, or talk to the team behind the platform.