The exchange PRD v2.0 retired the original Tier 0 (uncapped buying before a confirmed broker relationship) in favour of instant broker acceptance via API reliance. The substrate work, the same week, retired the standalone Level 1 account in favour of Level 1 auto-attestation under panel authority, with the revision published on the record. Two teams, working separately, walked back the same idea toward the same mechanism: instant, named, licensed broker acceptance, pre-authorised so it takes seconds. The February credit deck already assumed the third piece, CSCS-linked identity. This plan is not a merger of competing designs; it is the same design discovered from three directions.
Securities-backed lending on CSCS lien rails: LTV engine, margin management, capital-market BNPL. Turns held assets into working credit for 1.3M+ investors.
The market itself: ten product domains from investor app to matching engine, nine-ledger money architecture, clearing and settlement, surveillance, admin, issuer and partner APIs.
What everything stands on: one identity object with provenance, tiered broker attestation underwritten by the SEC caps, consent hooks enforced at the endpoint, the broker panel, the VBank cash rail, and in-jurisdiction inference for every model in a regulated decision path.
The exchange rule holds verbatim: no order reaches the market until minimum validated KYC and a named licensed broker acceptance exist. The substrate supplies how that happens in seconds: automated screening runs inline (an API call), the panel broker's Level 1 auto-attestation is the acceptance, and the manual-review tail proceeds at capped Level 1 instead of parking in limbo, with the SEC's own caps (₦20k a deposit · ₦200k cumulative) as everyone's bounded exposure and the exit ramp if review fails.
What each side contributed: the exchange PRD supplies the state names, the withdrawal gate and the hard rule; the substrate supplies the caps as legal risk budget (SEC three-tier, June 2017), the pre-authorised attestation that makes it instant, and Rule 67 as the licence category. Both documents' open questions on minimum KYC and broker reliance are hereby answered by the other document.
Both the exchange PRD and the credit deck name AI (pre-trade scoring, LTV offers, margin management) without resolving where inference runs. Resolved as follows: hot path (per-order surveillance scoring, margin monitoring) requires low-latency, in-jurisdiction inference close to the data, order flow and BVN-linked features do not leave the building; warm path (LTV offers, KYC anomaly) tolerates near-real-time on the same substrate; cold path (dashboards, insights) lives in the warehouse. Every inference is a read of its features and a write of its score on the provenance ledger (explainability = provenance for decisions); every automated hold shows honest state; every model has a deterministic lawful fallback, model-down never means check-skipped. The provider of the in-jurisdiction substrate is a governance decision taken openly, with related-party terms papered, alongside the cash rail.
The one-breath version for the room: we are one team building one platform in three layers; the corrections have already converged; the boundaries are contracts; the laws cross-cut; and the launch gate is already written, we just have to clear it together.