Agentic AI Risk & Controls Workshop

Build the controls the regulators expect for autonomous AI agents.

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.

Agent architecture diagram from patent materials

Day 1 — Understanding Agentic AI Risk

smart_toy
Agents vs. Traditional AI/ML

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

layers
Cumulative Operational Authority

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

psychology
Epistemic Drift

Why agents can execute valid logic on invalid premises.

show_chart
The Agency-Reliability Tradeoff

Agentic reliability is the product of three measurable dimensions — data sigma, process sigma, and agent sigma. Understanding their relationship is the key to improving it.

balance
MAS AIRG Agent Requirements

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

warning
Case Studies: What Can Go Wrong

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

Day 2 — Controls & Implementation

badge
Agent Identity & Inventory

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

shield
Runtime Governance

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

search
Reasoning Assurance

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

bar_chart
Six Sigma Measurement

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

bug_report
Testing Agents

Adversarial testing approaches, red teaming, failure scenario design.

admin_panel_settings
Human Oversight Models

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

route
Building Your Agentic AI Controls Roadmap

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

What You Get

  • slideshow
    Workshop materials — slides and reference guides
  • device_hub
    Agentic AI risk taxonomy customised to your institution's use cases
  • assignment_turned_in
    Draft agentic AI controls checklist aligned to AIRG
  • task_alt
    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

  • event
    2 days delivery + 2 days preparation = 4 advisory days.
  • group
    Can be delivered to multiple teams within the same institution at marginal additional cost.
  • handshake
    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 arrow_forward

Related Services

search
MAS AIRG Readiness Assessment

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

Learn More arrow_forward
device_hub
AI Governance Framework Design

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

Learn More arrow_forward
phone
AI Governance Retainer

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

Learn More arrow_forward