Canadian capital markets operations live with a short settlement cycle, heavy cross-border activity in securities listed in both Canada and the United States, and derivatives reporting under provincial rules. AI agents can chase unaffirmed trades, resolve routine instruction mismatches within limits and investigate cross-border breaks. People decide on costs and counterparty disputes. Provincial regulators and OSFI's model risk expectations apply.
What do capital markets operations leaders in Canada spend their time on?
Affirmation is the first race of the day. With a short settlement cycle, institutional trades must be confirmed and affirmed on trade date, and anything left over becomes a settlement risk. Operations chases investment managers and custodians for missing allocations, mismatched settlement instructions and late affirmations, often across time zones.
Cross-border activity adds the second load. Many large issuers are listed on exchanges in both countries, and clients buy in one market and sell in the other. Positions must move between depositories, currencies converted and corporate actions processed under two sets of market practices.
Third, reporting and controls. Derivative trades must be reported under provincial rules, and dealers must meet the national self-regulatory organisation's requirements on books, records and client asset segregation. Every system change in the trade flow must be introduced without breaking settlement.
Where can AI agents help?
- Affirmation chasing (depth 3). The agent watches unaffirmed trades through the day, sends standard chasers to the right contacts at the counterparty, and records responses. Where a counterparty asks for a change to economic terms, the agent holds the trade for a person.
- Instruction mismatch repair (depth 3). For mismatches where the counterparty's standing instructions on file are clearly current, the agent corrects the instruction and re-matches within limits. New or changed standing instructions received by email are held, since they are a fraud route.
- Cross-border break investigation (depth 2). For positions that did not move between depositories, the agent reads both sides' records and the related FX, identifies the cause and drafts the fix for a person.
- Reporting reconciliation (depth 2). The agent compares derivative reports with the trading book and drafts corrections for review.
Cross-border work needs particular care with sources. The Canadian and US depositories each hold a sealed record of the position; a counterparty's email saying shares were delivered is external until one of them confirms it. The agent should report breaks against the depository records and never close a break on the strength of a message alone.
Rollout should be gradual. Starting with one desk or one counterparty group, then widening as evidence builds, limits the damage if the agent misbehaves in live conditions. That is the logic of a canary release, applied to an agent's authority rather than to code. The broader method is in deploying AI agents: train, shadow, certify, widen.
What stays with people?
Changes to the economic terms of a trade, acceptance of new standing settlement instructions, and decisions to buy in or claim against a counterparty stay with people. So do decisions about who bears the cost of a fail.
Responses to regulators and the self-regulatory organisation, breach assessments and any finding that client assets were not properly segregated stay human.
People also decide when an agent is no longer needed. Processes change, and an agent left running against a retired workflow can do damage. Our AI agent decommissioning checklist covers how to switch one off cleanly, revoke its access and keep its records.
Which rules apply in Canada?
Provincial securities regulators set and enforce securities law, including rules on institutional trade matching, derivatives trade reporting and dealer conduct. A national self-regulatory organisation oversees investment dealers, including their books and records, client asset protection and operational controls.
Where the dealer is part of a federally regulated bank, OSFI supervises the group. OSFI Guideline E-23 on model risk management sets expectations for governing models through their life cycle, which reaches an agent that makes or informs operational decisions: inventory, validation, monitoring and decommissioning. OSFI also expects sound management of third-party risk and operational resilience.
PIPEDA governs personal information about clients and counterparty contacts. There is no federal AI act in force; the NIST AI Risk Management Framework and ISO/IEC 42001 are useful references.
This is an explanation of the rules, not legal advice; confirm how they apply to you with counsel.
How do you start in the first 90 days?
Days 1 to 30: choose affirmation chasing. The agent sends standard messages and records responses, which is low-risk and frees people for real problems. Register it in the model inventory and agree validation.
Days 31 to 60: shadow the agent. It proposes chasers and logs what it would have done; your team works as normal. Compare timing and accuracy.
Days 61 to 90: certify at depth 3 for one counterparty group, then widen group by group as a canary. Begin shadowing instruction mismatch repair. India's capital markets operations face different structures; see our brief on operations agents in Indian capital markets.
Questions people ask
Can the agent affirm trades on the dealer's behalf?
Affirmation is often done by the investment manager or custodian. The agent chases and records; where your firm affirms, it may do so within certified bounds for matched trades.
How does the agent handle time zones?
It works the queue continuously and knows each counterparty's hours. Messages outside those hours are queued, not sent.
What does E-23 expect when an agent is retired?
Model life cycle management includes decommissioning. Record the decision, revoke access, retain records and confirm nothing depends on the agent.