AI & Data Assessments
What assessments does DAVHILL Group provide?
DAVHILL Group provides comprehensive AI, data, and governance assessment services that help organizations adopt advanced technologies responsibly and with confidence. Our assessments evaluate AI readiness, data quality, governance maturity, risk exposure, model assurance, security, and operational feasibility—giving leaders a clear, actionable understanding of their current state and the steps required to move forward safely. With deep expertise in public‑sector environments, we help clients strengthen oversight, improve data reliability, meet Government of Canada requirements, and ensure AI systems are transparent, trustworthy, and aligned with mission and policy objectives.
What is an Algorithmic Impact Assessment (AIA)?
An Algorithmic Impact Assessment (AIA) typically determines the risk impact of an automated decision algorithm on stakeholders, organizations or the general public.
The automated decision in question does not necessarily need to be part of an AI or ML system, any non “human-in-the-loop” (HITL), system should be evaluated depending on the level of risk.
As a baseline we perform our assessments in line with the Government of Canada (GoC) Algorithmic Impact Assessment (AIA) (part of the Canadian Directive on Automated Decision-Making). We add in additional controls, checkpoints and references to ensure that the feedback that you get from our assessors enables you to meet industry and government best practices, including but not limited to OECD AI Principles, ISO/IEC 42001:2023 Artificial Intelligence Management System, NIST AI Framework, Law Commission of Ontario Human Rights AI Impact Assessment, and other related best practices such as the GoC Privacy Impact Assessment (PIA).
How do you rate an Algorithmic Impact Assessment (AIA)?
Our AIA evaluation is scored to the following categories:
Risk: Project, System, Algorithm, Decision, Impact, Data
Risk Mitigation: Consultations, De-risking and mitigation measures
Each category is given a risk score between 1 (little to no impact) up to 4 (very high impact) which is then weighted and aggregated for an overall score.
The GoC Directive on Automated Decision Making requires that any project that is rated above a Level 2 requires a third party peer review (Appendix C - Impact Level Requirements). Professionals at DAVHILL can perform this peer review, or if internal resources or timing is restricted also the initial assessment.
What is an AI Readiness assessment?
An AI Readiness assessment evaluates whether an organization has the foundational capabilities required to adopt AI responsibly and effectively. It examines data maturity, infrastructure, governance, security posture, and workforce capability to determine whether the environment can support AI systems without introducing operational, ethical, or compliance risks. The assessment highlights where the organization is prepared for AI and where structural gaps must be addressed before moving forward.
For Government of Canada and environments, readiness extends beyond technology. It includes alignment with policy obligations, clarity of roles, procurement constraints, and the ability to maintain oversight and mission assurance. The assessment provides a prioritized roadmap that identifies blockers, strengthens governance, and ensures AI adoption improves — rather than complicates — organizational performance.
What is a Data Quality & Maturity assessment?
A Data Quality & Maturity assessment evaluates the reliability, completeness, and usability of data across the organization and is a key component of good data governance. It examines data quality dimensions (accessibility, timelines, relevance, consistency, accuracy, completeness, interdependency, uniqueness) and assesses how data is collected, governed, transformed, and consumed. The goal is to identify where data issues originate, how they propagate, and how they impact reporting, analytics, and AI systems.
For public‑sector environments, the assessment also considers bilingual data, fiscal calendars, legacy systems, and cross‑departmental interoperability issues. The output is a clear, actionable view of data strengths and weaknesses, along with recommendations to improve data pipelines, stewardship, metadata, and governance so that AI systems have the reliable “fuel” they require.
What is a Model Validation & Assurance assessment?
A Model Validation & Assurance assessment provides a deep review of model performance, robustness, fairness, and reliability. It evaluates how models are trained, tested, monitored, and documented, and whether they meet standards for accuracy, stability, and “explainability”. It also examines drift detection, versioning, auditability, and reproducibility — all essential for high‑stakes environments.
The assessment ensures that models behave as expected, remain stable over time, and can be defended to auditors, executives, and oversight bodies. It is particularly valuable for departments deploying predictive models, risk scoring, or decision‑support systems that require defence‑grade assurance. This assessment requires good data science practices inside the organization that is being assessed.
What is a Data Pipeline & Infrastructure assessment?
A Data Pipeline & Infrastructure assessment evaluates the health, reliability, and scalability of data pipelines and infrastructure. It examines ingestion, transformation, orchestration, storage, cloud architecture, interoperability, and security to determine whether the technical environment can support modern AI workloads without introducing fragility or operational risk. This assessment required good data engineering practices inside the organization that is being assessed.
For GoC clients, the assessment also considers sovereignty, multi‑cloud constraints, legacy systems, and integration with enterprise platforms. The output is a clear roadmap for strengthening infrastructure so AI systems can operate securely, efficiently, and at scale.
What is an AI Lifecycle & Project Management assessment?
An AI Lifecycle & Project Management assessment examines whether teams have the processes, controls, and discipline required to deliver AI projects safely and effectively. It evaluates lifecycle stages, documentation, change management, versioning, peer review, and governance checkpoints.
The result is a clear view of whether the organization can manage AI projects with the rigour required in federal and defence environments — and a roadmap for strengthening project governance, delivery practices, and oversight.