Ducaltus | AI Assurance & Governance

AI assurance extends beyond model evaluation.

Strong model performance alone does not establish that an AI system is ready for deployment or that its assurance position will remain unchanged as conditions evolve. Ducaltus provides an operational assurance layer connecting evaluation evidence with deployment readiness, governance oversight, remediation, reassessment, and ongoing assurance across the AI deployment lifecycle.

Who We Support

For organisations deploying AI where assurance matters.

Ducaltus supports organisations developing, evaluating, procuring, or deploying AI systems in high-stakes and governance-sensitive environments. It is designed for contexts where deployment decisions require defensible evidence, structured oversight, and continuing assurance beyond initial model evaluation.

AI Developers & Deployers

Teams moving AI systems from technical evaluation toward controlled operational deployment and continuing assurance.

Public-Sector & Regulated Organisations

Organisations requiring structured assurance, governance oversight, and defensible evidence around consequential AI deployment.

AI Governance & Risk Teams

Functions responsible for AI risk, oversight, accountability, deployment review, and continuing governance.

Independent Evaluators & Assurance Partners

Organisations conducting technical, governance, procurement, or third-party assurance of AI systems.

Capabilities

Operational assurance across the AI deployment lifecycle.

Deployment Readiness & Risk

Structured evaluation of whether available evidence supports deployment, restriction, further review, or reassessment.

Subgroup Reliability & Fairness

Evaluation of performance variation, error disparities, fairness concerns, and hidden reliability failures beyond aggregate metrics.

Threshold & Error Trade-offs

Analysis of threshold-sensitive behaviour and false-positive and false-negative trade-offs under relevant deployment conditions.

Governance & Oversight

Structured support for governance review, accountability, deployment oversight, and evidence-informed decision-making.

Remediation & Reassessment

Support for tracking assurance concerns, remediation activity, reassessment, and subsequent deployment review.

Assurance Evidence & Traceability

Persistent assurance evidence designed to support review, justification, traceability, and audit reconstruction.

Ducaltus Platform

From evaluation evidence to governed deployment decisions.

Ducaltus provides a structured assurance environment for connecting technical evaluation with operational review, governance decisions, remediation, reassessment, and continuing evidence across the deployment lifecycle.

Evidence-Driven Assurance

Bring evaluation evidence and assurance findings into a structured operational review process rather than relying on isolated metrics or one-time assessments.

Lifecycle Governance

Support assurance decisions before deployment and through subsequent review, remediation, reassessment, and changing operational conditions.

Human Oversight

Preserve accountable human review and governance authority around consequential deployment and assurance decisions.

Traceable Decision Evidence

Maintain evidence and decision context that can support internal review, governance justification, and audit reconstruction.

Assurance Approach

From evidence review to continuing assurance.

Ducaltus applies a structured assurance process that connects technical evaluation with deployment context, governance review, remediation, reassessment, and continuing evidence.

1. Define the Assurance Context

Establish the AI system, intended use, deployment conditions, available evidence, relevant stakeholders, and assurance objectives.

2. Evaluate Evidence & Reliability

Review model behaviour, subgroup reliability, error trade-offs, threshold sensitivity, and other deployment-relevant evidence.

3. Assess Deployment Readiness

Examine whether the available evidence supports deployment and identify conditions requiring restriction, further review, or additional assurance activity.

4. Support Governance & Remediation

Record assurance findings, support accountable review, and track remediation or other actions arising from identified concerns.

5. Reassess & Maintain Assurance

Re-evaluate relevant evidence following remediation or material change and maintain traceable assurance information over time.

Research & Evidence Base

Research informing the Ducaltus assurance approach.

Ducaltus draws on research spanning AI reliability, fairness metric disagreement, subgroup performance, threshold sensitivity, deployment risk, and operational assurance. This research provides part of the evidence base informing the company's approach to high-stakes AI assurance and governance.

Operational AI Deployment Assurance Research extending decision-aware fairness and deployment-risk evaluation toward operational AI assurance infrastructure, governance-state orchestration, remediation-aware deployment control, and threshold-sensitive deployment governance. View research When Fairness Metrics Disagree Research introducing the Fairness Disagreement Index (FDI) for analysing disagreement between fairness metrics and its implications for AI reliability assessment. View research FairRisk-FDI Decision-aware deployment risk evaluation connecting fairness disagreement, threshold instability, subgroup reliability, and operational deployment uncertainty in high-stakes AI systems. View research IFEM: Intersectional Fairness Evaluation Research on intersectional subgroup reliability, threshold-sensitive fairness behaviour, and deployment-relevant subgroup instability in face recognition systems. View research Reliability & Risk in AI-Assisted Medication Systems Research exploring reliability, decision risk, and failure modes in AI-assisted systems operating in sensitive decision environments. View research Fairness in Law Enforcement Facial Recognition Research focused on fairness, deployment considerations, and subgroup performance analysis in high-impact facial recognition settings. View research

Research to Operational Assurance

Translating research into practical assurance infrastructure.

Ducaltus connects research on reliability, fairness, deployment risk, and governance with practical assurance processes for evaluating and governing high-stakes AI systems across the deployment lifecycle.

Research Notes & Insights

Short perspectives on AI assurance, fairness, and deployment risk.

Technical perspectives exploring subgroup reliability, fairness evaluation, deployment risk, and operational considerations in high-stakes AI systems.

Why Aggregate Accuracy Fails in High-Stakes AI

Strong overall accuracy can conceal serious subgroup failures and hidden deployment risks in operational AI systems.

Read Insight

When Fairness Metrics Disagree

Different fairness metrics can produce conflicting conclusions about the same model, creating uncertainty in deployment decisions.

Read Insight

False Positives vs False Negatives

The operational impact of AI system errors depends heavily on context, deployment conditions, and real-world consequences.

Read Insight

Explore all insights →

Founder

Khalid Adnan Alsayed

Founder of Ducaltus, working across operational AI assurance, deployment governance, AI reliability, fairness, and the translation of technical evidence into accountable governance for high-stakes AI systems.

BSc (Hons) Artificial Intelligence · Teesside University MRes Artificial Intelligence & Public Policy Candidate · University of Leicester

Assurance requires evidence, not assumptions.

Ducaltus connects technical evidence with deployment governance, accountable review, remediation, reassessment, and traceable assurance to support defensible decisions across the AI deployment lifecycle.

Evaluation & Engagement

Explore Ducaltus for your AI assurance requirements.

Ducaltus is available for evaluation discussions, controlled pilot engagements, technical and governance review, and strategic collaboration relating to high-stakes AI assurance.

Initial discussions can begin using public, non-confidential material. More detailed technical and operational documentation is available to appropriate evaluators and partners under confidentiality arrangements.

Discuss an evaluation View capability statement