Answer — AI Governance
Who is the accountable person for AI systems under the AMF guideline?
The AMF’s AI guideline, in force May 1, 2027, expects senior management to designate a person accountable for all of the institution’s AI systems (SIA). The choice belongs to the institution; the expectation is a clear designation — someone able to answer for the AIS inventory, the risk ratings, and ongoing monitoring.
What the guideline expects
The AMF’s guideline on the use of artificial intelligence — final since April 2026, in force May 1, 2027 for Québec-authorized insurers, financial services cooperatives, trust companies, and authorized deposit institutions — places AI governance at the level of senior management. The expectation is plain: senior management should designate an accountable person (« personne imputable ») for all of the institution’s artificial intelligence systems (AIS, « SIA »).
The guideline’s other expectations orbit that designation: a centralized AIS inventory listing the AIS whose risk is non-negligible, a risk rating updated periodically, controls at every stage of the AIS lifecycle, and ongoing monitoring. The vocabulary is that of a guideline — expectations (“should”), not statutory requirements — but it is the text the regulator will rely on when it examines the institutions it oversees. The binding text is French.
Who does that mean, concretely?
The choice belongs to the institution: the guideline expects a designation by senior management and leaves the org chart to the institution. The useful question is therefore not “which title?” but “who can answer?”. The accountable person will need to answer, at any moment, very concrete questions: which AIS are in use, including the ones that arrived through a vendor; which carry non-negligible risk, and with what rating; where each system stands in its lifecycle; and what ongoing monitoring has found, and what was then corrected.
Designating a person without giving them the means to answer creates accountability without creating capability. The most demanding designation criterion is not hierarchical: it is access to a picture of the estate that stays true between two examinations.
Can a platform carry that accountability?
No — and any offer that implies it deserves a careful look. Accountability cannot be delegated to a tool: that is precisely the point of the designation the AMF expects. What a tool can carry is the role’s work: keeping the inventory current as the environment changes, recording ratings and their revisions, timestamping what ongoing monitoring finds, generating the extracts an examiner will ask for. The accountable person remains the one who answers.
That is the division Agentica applies. The system of record continuously maps every AI agent in your business environment and records it in a tamper-proof history — read-only, metadata and signals only. The accountable person finds there the substance of their answers: the living inventory, dated gaps, recorded corrections. They remain accountable; they stop being alone in front of the question.