Portfolio intelligence built for
evidence before action.
Nexum Fiducia is the institutional expression of the Nexum architecture: deterministic portfolio artifacts, review-gated document drafts, and a strict API contract designed for private banks, family offices, RIAs, and fiduciary teams.
What the implemented engine proves today
Deterministic Portfolio Evidence
A synthetic-data engine produces sealed forecast, constrained-allocation, trade, recommendation, and four-dimension CRI artifacts.
Review-Gated Drafts
Fiducia-only workflows generate supplied-facts IPS, Form ADV Part 2A, and ERISA fiduciary-process drafts with source and review boundaries.
Replayable Audit Contracts
Hash-linked artifacts support deterministic replay, signed audit records, and configured Prufs event delivery with strict receipt validation.
One ordered evidence chain,
with clear ownership boundaries.
The implemented Nexum Engine separates evidence construction from the institutional controls that remain the product consumer's responsibility.
Forecast Artifact
A deterministic Random Forest forecast over an entirely synthetic factor fixture, sealed with model, data, training, background, and explainer hashes.
Constrained Allocation
Long-only optimization exposes expected-return and covariance inputs, constraint residuals, binding bounds, sensitivities, and counterfactual solves.
Trade Explanation
Fractional-share rebalance evidence reconciles starting holdings and cash to target weights, turnover, ending cash, and total value.
Recommendation Bundle
A hash-linked manifest preserves forecast context, allocation basis, and trade implementation without claiming one explanation substitutes for another.
Regulated Drafts
IPS, Form ADV Part 2A, and ERISA process documents are supplied-facts drafts that remain blocked from filing or execution pending authorized review.
Consumer Contract
A strict authenticated API handshake pins product identity, entitlements, operation sequence, engine version, and the exact OpenAPI digest.
What integration still has to provide
- Authenticated end-user sessions, role mapping, tenancy, and account persistence
- Validated client facts, mandate collection, suitability, fiduciary judgment, and legal review
- Real-market and portfolio data, monitoring, rate limiting, and operational key management
- Broker or custodian execution, durable product-edge workflows, and production acceptance
Explain each decision at
the layer that made it.
Nexum does not stretch a forecast explanation into a claim about allocations or trades. Each layer emits its own typed evidence.
Designed for teams that must
show their work.
Nexum Fiducia is being designed around the evidence and review needs of fiduciary organizations. Institutional discovery will shape the consumer workflow that sits above the engine.
Private Banks
Explore auditable portfolio evidence for high-net-worth and ultra-high-net-worth mandates.
Family Offices
Evaluate consistent evidence and review boundaries across complex investment operations.
Registered Investment Advisers
Assess how deterministic artifacts could support adviser review without displacing adviser judgment.
Trust Departments
Examine supplied-facts documentation and ordered review workflows for fiduciary processes.
A useful first conversation
- Which mandate and approval constraints must be machine-verifiable?
- Which recommendation evidence must reach portfolio, compliance, and client reviewers?
- Where must source facts, approvals, and execution authority remain human-controlled?
- Which audit receipts and replay checks would your examiners expect to inspect?
Institutional evaluation,
scoped before it is priced.
Nexum Fiducia is not generally available and has no public production license. Evaluation scope depends on the institution, mandate, data boundary, and required integration evidence.
Four decades of AI,
from econometrics to agents.
Wade Lovell is principal of CognitionHive. His AI journey began in a 1981 doctoral seminar in Econometrics at the College of William & Mary. Today he builds measurement and evidence frameworks for agentic AI in regulated settings.
Across CTO, Technical Architect, and AI Architect roles in banking, insurance, and enterprise software, Wade has led multi-million-dollar technology transformations for Fortune 500 clients. His work at Salesforce placed him at the intersection of enterprise architecture, data strategy, and agentic AI.
His doctoral research at Walsh College develops the Claim Reliability Index: four dimensions—Correctness, Faithfulness, Stability, and Constraint Compliance—for evaluating implemented engineering gates around agent outputs.
Wade is a published author on ethical AI in education and is developing practical product architectures that keep evidence, review, and human authority visible.
Start with the mandate,
the controls, and the evidence.
Tell us what your institution would need to evaluate. This form opens a discovery conversation; it does not enroll you in a production service.