Our Offering
Systems for Trust & Agency — Engineering Legitimate AI in the Age of Intelligent Infrastructure
The Architecture for Trust & Agency
Three organisations form an integrated architecture spanning foundation, system, and execution — together defining, encoding, and embedding trust in AI-driven organisations.
Explores sovereignty, governance, identity, reciprocity, and institutional design. Reframes trust as a systems property shaped by deep design decisions. Trust must be defined structurally before it can be implemented operationally.
Transforms fairness and governance principles into structured capability frameworks. Embeds fairness objectives, intervention strategies, regulatory alignment, and monitoring logic into the AI lifecycle. Trust becomes measurable and scalable.
Partners with executive teams to redesign AI-enabled operating models. Aligns governance, analytics, product development, and leadership accountability. Trust becomes anchored in institutional practice.
vfairness Library
The most complete open-source ML fairness toolkit available — with a commercial Pro edition for enterprise workflows, compliance, and white-label output.
vfairness
Free — MIT License
pip install vfairness
vfairness-pro
Commercial License
pip install vfairness-pro
Digital trust is not a free lunch. It takes work and it must be earned. We make it within reach.
Validant AI Fairness Platform
A hosted SaaS platform for regulated enterprises, audit firms, and legal/compliance teams — interactive tools that go beyond what a library can offer.
Causal DAG Builder
Visual directed acyclic graph editor. Drag-and-drop variables, draw causal edges, auto-generate from proxy analysis, and compute path-specific fairness effects.
Decision Flip Testing
Automated perturbation of protected attributes with flip rate computation, severity scoring, SHAP/LIME attribution, and API endpoint probing for black-box models.
Counterfactual Suite
Twin test framework, counterfactual data generation from causal DAGs, group-level counterfactual analysis, and exportable explanation cards.
10-Phase Assessment Wizard
Step-by-step guided workflow automating the full EU AI Act compliance audit — from data validation through to model card generation.
What-If Simulator
Multi-dimensional real-time exploration of thresholds, constraints, and calibration methods. Pareto frontier search and shareable stakeholder views.
Evidence Packages
One-click export of the entire assessment — SVGs, metrics, DAGs, audit trail, model card — zipped and cryptographically signed for regulatory submission.
Why This Ecosystem
What you gain by building on the validant.ai, iceberg.digital, and Glinz & Company trust architecture.
Full-pipeline fairness from day one
vfairness covers detection, mitigation, calibration, and monitoring in a single MIT-licensed library. No stitching together five different tools — install once and run a complete audit from the command line.
Regulatory readiness built in
Pre-built templates for the EU AI Act, NYC Local Law 144, and EEOC guidelines mean your compliance team can produce audit-grade evidence packages without starting from scratch.
Trust grounded in theory, not just tooling
iceberg.digital’s research on sovereignty, governance, and institutional design gives you a structural framework for trust — not just metrics, but a defensible logic for why your system is fair.
Enterprise execution, not just advice
Glinz & Company embeds fairness into your actual operating model — aligning leadership, product development, analytics, and compliance into a single accountable structure.