Regulatory analysis, written from inside the artifact.
Long-form notes on the frameworks, enforcement actions and operating-model questions that land in front of us every week. Written for practitioners in the second line of defence, for the risk committees that read their reports, and for the counterparties reading yours.
The three pieces we keep sending to clients.
Each of these lands on a question that has moved from academic to invoiced over the last six months, AI liability, AI-insurance binding, and the collision between legacy data governance and AI governance programmes.
AI Agent Insurance: No Governance Evidence, No Policy
The AI-liability underwriting questionnaire has started to mirror what a supervisor would ask. Without a portable evidence pack, the bind does not happen and the renewal does not price.
Read the note →Liability · EU PLDAI Product Liability in Financial Services
The revised EU Product Liability Directive reclassifies software, including AI, as a product. Burden of proof shifts toward the manufacturer. Canadian and US exporters inherit the risk.
Read the note →Data · AI GovernanceData Governance, Reshape for the AI Era
Running a legacy data governance programme and an AI governance programme in parallel is the single largest hidden cost inside regulated institutions. One artifact set should answer both.
Read the note →Additional reading
The Five AI Governance Artifacts Every Canadian FSI Needs
The core artifact set a Canadian FSI supervisor will recognise, inventory, model cards, validation, monitoring, and the control library beneath them.
Canadian FSIFINTRAC 2026 Amendments and CIRO Consolidation
What the FINTRAC amendments and CIRO consolidated rulebook mean for AI deployed inside Canadian AML, KYC and trade supervision.
MonitoringThe Intervenability Turn, Monitoring Isn't Enough in 2026
Regulators are moving past detection and into the intervenability question, can you stop, throttle or unwind the AI before harm compounds?
Vendor · B-10OSFI E-23 Vendor Cascade
How the E-23 obligation reaches into the vendor stack and why nth-party AI discovery is now a table-stakes diligence artifact.
Fintech · VendorThe OSFI E-23 Vendor Playbook
How fintechs selling into Canadian Schedule I banks answer AI-governance questionnaires without re-papering every engagement.
OSFI · FIFAIWhat OSFI Is Signalling Through Its AI Governance Workshops
FIFAI is a channel, not a rulebook. How OSFI's supervisory themes cascade through E-23, B-10 and E-21 to shape AI governance programmes in Canadian FRFIs.
Operating modelAI Governance Consultancy or Platform, When to Choose Each
The build-vs-buy decision for AI governance infrastructure. Where platforms win, where specialists win, and why the answer is almost always both.
Cross-border MRMSR 11-7 Meets OSFI E-23, A Cross-Border Crosswalk
For firms operating across US and Canadian supervisory perimeters, one validation file that answers both, and the gotchas in the mapping.
Operating modelThree AI Governance Gaps That Existing Tools Cannot Close
Where the current generation of GRC and MLOps tooling stops and what an AI-era governance operating model has to carry.
The reading is the opening act. The operating model is the work.
Every piece above maps to a playbook, a framework and a control set the firm runs. Start with an enquiry; we'll walk you from the note to the artifact.