Agentic AI Risk & Controls Workshop

The AIRG requirement your team doesn't know how to handle.

What It Is

A focused 2-day workshop for AI teams, risk teams, and technology leadership on the specific risks introduced by AI agents — autonomous AI systems with tool access and decision-making authority — and the controls MAS expects. This addresses the part of AIRG that most financial institutions are least prepared for.

The AIRG explicitly identifies AI agents as introducing heightened risk, recognising that agents with tool access can autonomously execute actions with real-world consequences beyond those of traditional predictive models.

Day 1 — Understanding Agentic AI Risk

Agents vs. Traditional AI/ML

What makes AI agents fundamentally different from traditional AI/ML models.

Cumulative Operational Authority

Why an agent's effective permissions are emergent, not static.

Epistemic Drift

Why agents can execute valid logic on invalid premises.

The Agency-Reliability Tradeoff

Current agentic processes operate at 1–1.5 sigma — below the quality threshold for any mission-critical process.

MAS AIRG Agent Requirements

MAS AIRG specific requirements for AI agents: failure modes, enhanced testing, human oversight.

Case Studies: What Can Go Wrong

Trading agents, customer service agents, underwriting agents — real-world failure scenarios examined.

Day 2 — Controls & Implementation

Agent Identity & Inventory

How to register and profile autonomous agents using Corvair's ten-layer governance model.

Runtime Governance

Just-in-time privilege, zero standing access, emergency kill switches.

Reasoning Assurance

How to audit whether an agent's reasoning chain is valid — using composable lenses, decision validity warrants, and the SCAR scoring rubric.

Six Sigma Measurement

Defining sigma targets for agentic workflows, measuring data sigma and process sigma, applying DMAIC.

Testing Agents

Adversarial testing approaches, red teaming, failure scenario design.

Human Oversight Models

In-the-loop vs. on-the-loop vs. out-of-the-loop — when each applies.

Building Your Agentic AI Controls Roadmap

Participants leave with an action plan tailored to their institution's agentic AI deployments.

What You Get

  • Workshop materials — slides and reference guides
  • Agentic AI risk taxonomy customised to your institution's use cases
  • Draft agentic AI controls checklist aligned to AIRG
  • Participant action plans

How It's Different

No other consultant in Singapore can walk into a bank and teach agentic AI governance from their own original methodology. Cumulative operational authority, epistemic drift detection, and composable auditable reasoning are the curriculum — not synthesised from other people's research, but built from first principles and protected by patent filings.

Engagement Details

  • 2 days delivery + 2 days preparation = 4 advisory days.
  • Can be delivered to multiple teams within the same institution at marginal additional cost.
  • Workshop content can also be licensed to systems integrators.

Equip Your Team for Agentic AI Governance

Give your AI, risk, and technology teams the knowledge they need to govern autonomous agents — taught by the practitioner who developed the methodology.

Schedule a Briefing

Related Services

MAS AIRG Readiness Assessment

A structured evaluation of your institution's current AI governance posture against MAS AIRG requirements.

Learn More
AI Governance Framework Design

A tailored governance framework that addresses your institution's specific regulatory and operational requirements.

Learn More
AI Governance Retainer

Ongoing advisory support for institutions that need continuous access to agentic AI governance expertise.

Learn More