AI Readiness & Governance

The federal Directive on Automated Decision-Making requires departments to complete an Algorithmic Impact Assessment before deploying any system that supports or replaces administrative decisions. We guide your team through the AIA process, help determine the impact level (Level 1 through 4), document the required controls for your level, and prepare the materials for publication on the Open Government Portal. We also help interpret how the Directive applies to your specific use case, which is often the hardest part.

Every organization adopting AI needs internal policies governing how it's used, by whom, and under what conditions. We develop acceptable use policies, generative AI guidelines aligned with the TBS FASTER principles (Fair, Accountable, Secure, Transparent, Educated, Relevant), risk classification frameworks, and approval workflows. For organizations beyond federal government, we tailor governance documentation to your sector's requirements, whether that's provincial directives, OSFI guidance for financial services, or board-level governance for non-profits and enterprises.

Most organizations either have too many AI ideas with no way to evaluate them, or they can't articulate where AI would actually help. We run structured workshops with your teams to identify practical AI use cases, then score each one on feasibility, risk, expected impact, data readiness, and alignment with your strategic priorities. You get a ranked backlog of AI opportunities with clear next steps for the top candidates, not a wish list, but a plan.

The AI vendor market is noisy and moves fast. We help you evaluate AI tools and platforms using criteria that matter for Canadian organizations: data residency and sovereignty, PIPEDA compliance, alignment with TBS guidelines, integration with your existing systems, and total cost of ownership. We apply the same structured evaluation approach we use in our Procurement Advisory work, adapted for the specific risks and considerations that AI tools introduce, including model transparency, bias risk, and vendor lock-in.