I wrote the AIRG while leading AI risk supervision at the Monetary Authority of Singapore, after almost two decades in risk management and writing regulations. On the newer end of this, I co-authored a paper on runtime governance for agentic AI and helped a little with MAS's SAFR, the industry's reference for safeguarding agentic finance at runtime. This guide is for the people who own the adoption decision: the Board and Senior Management who set the appetite, approve the spend, and answer for the outcome. It works through the decisions you actually make, and what the AIRG and the Project MindForge toolkit cover for each one.

The through-line is simple. Governance and adoption are not in tension. Rapid, useful adoption of AI is what good governance buys you: you move faster, not slower, when you can trust that a system will behave, follow the law, and be caught when it does not. The job at your level is to make adoption responsible so that it can scale.

How it fits together. The FEAT principles, fairness, ethics, accountability and transparency, are the underlying principles. The AIRG is the expectations built on them. MindForge is the practice, based on and aligned to the AIRG. Principles, then expectations, then practice.

I keep this openly available rather than behind a sign-up. The people who have to decide on AI should understand it well, and the system works better when they do. Use it, and pass it on. And if you would like to follow me for more of such stuff, come find me on my Substack.

Each topic opens with a question you will recognise from your own meetings, and each one links to the fuller write-ups on this site and to the matching ebook. Use the contents below to jump to a decision, or step through them in order. The companion guide for the people who oversee and challenge AI, the second and third line, is here.

I am keeping this as a live document, so drop me a note (at gary@quaintitative.com) anytime if you want to add or correct something. Your feedback helps improve the guide for everyone.

What this covers, and what else is here

These six decisions are a doorway, not the whole house. Behind them sits the AIRG in full and all seventeen MindForge considerations, written up one by one, plus the wider Singapore picture, comparisons with the global frameworks, and the tools you need to actually use any of it. A few of the highlights:

The full MindForge toolkit, consideration by consideration. The decisions here draw on key considerations. The rest are written up in the MindForge toolkit.

The companion guide for your control functions. The same material for the people who oversee, challenge and assure AI, the second and third line: AI risk management and governance for risk and control functions.

Crosswalks that map one framework to another. The AIRG to MindForge crosswalk, clause by clause; the AIRG, MindForge and the CRI framework together; and the AIRG to FSB crosswalk.

How the AIRG compares with the global frameworks. The AIRG compared with the EU AI Act, NIST and ISO 42001, so a firm across jurisdictions maps once.

The resource library and the ebooks. Everything in one place: AI risk management in finance, the resources, and the plain-English books.

Contents

The six decisions

Decision 1 of 6

What is changing, and which of it is yours to own

The landscape, and where the Board and Senior Management sit in it.

Generative and agentic AI are moving fast. Which parts of this are actually the Board's and Senior Management's to own?

What is changing. Generative AI produces open-ended content and leans heavily on third-party models. Agentic AI goes further: it plans, calls tools and takes actions across steps, with less human involvement at each one. Both open real value in finance, in service, research, operations and compliance, and both raise the stakes, because capability and risk rise together. The decision at your level is not which tool to use; it is how far and how fast to adopt, and on what terms.

MindForge names five roles. The toolkit splits the firm into five: Executives, Builders, Custodians, Use Case Owners, and Business Users. Builders make the AI, Use Case Owners are accountable for a given use, Custodians are the oversight and control functions, Business Users apply it, and Executives, you, set the direction and carry the accountability. Knowing who holds what keeps the Board from either rubber-stamping or trying to do the first line's job.

What you own. The AIRG and MindForge put a short, non-delegable list on the Board and Senior Management: approve the overall approach, set the risk appetite for AI, make sure roles and accountability are clear across the three lines, understand enough to challenge what you are told, and keep that up as the technology changes. You do not build or run the models. You own whether the firm is fit to adopt them.

What good looks like, and what to ask. A Board that can say, in plain terms, where AI is used, how much risk that carries, and who answers for it. Ask for the one-page picture of the firm's AI and its materiality, and ask which decisions on this list the Board has actually taken, rather than noted.

In an ebook: Agentic and Generative AI for Finance.

Decision 2 of 6

Should we build this, and what will governing it cost

Proportionate adoption: materiality first, then the true cost of running it safely.

A business case lands on your desk for a new AI use. What should you weigh before approving it?

Start with how material it is. Before the value, ask how much this use could affect: its impact on the firm and its customers, how complex or novel it is, and how far it replaces human judgement. The AIRG works by materiality, and MindForge tiers uses, commonly low, medium and high, on exactly these lines. The tier is not paperwork; it decides how much governance the use needs, and therefore what it will cost to run safely.

Count the governance cost, not just the build cost. A high-materiality use carries a heavier control load for its whole life: independent review before go-live, monitoring after, human oversight, and the people to do all three. A business case that counts only the build and licence cost understates what the firm is signing up for. The right question is whether the value still holds once the cost of governing the use to its tier is in the sum. Sometimes a lighter, lower-materiality design captures most of the value for a fraction of the control load; that is a decision worth forcing.

Check the firm can actually host it. Infrastructure fitness is part of the decision: can your environment run, secure, log and monitor this use to the standard its tier demands, and keep sensitive data where it belongs? If it cannot, either the infrastructure is part of the cost, or the use belongs with a third party, which brings its own controls.

For bought AI, decide what to demand. Where the AI comes from a vendor, the disclosure you require, an AI Card, model or system card, scales with how material your use is. Buying it does not move the accountability to the vendor. What you cannot get disclosed, you cover with your own testing or contract terms.

What good looks like, and what to ask. A business case that states the use's materiality tier and the cost of governing it to that tier, not value alone. Ask: what tier is this, what does governing it to that tier cost over its life, can we host it safely, and does the value still hold with all of that counted?

In an ebook: Scaling AI Governance.

Decision 3 of 6

Is the organisation set up to run AI at scale

Responsibilities, risk appetite, the risk taxonomy, the inventory and the dashboard.

The firm wants to scale AI from a handful of uses to many. Is the organisation set up to carry that safely?

Set the appetite, and the responsibilities. The Board approves the firm's approach to AI and sets how much AI risk it is willing to carry, and against which measures. Senior Management runs the framework the Board approves. The AIRG does not prescribe the structure, but it requires clear accountability and no gaps, across the three lines of defence.

Choose how AI risk sits in the firm's risk taxonomy. MindForge sets out three approaches, which can be combined: update each existing risk category for AI (distributed), expand model or technology risk to hold AI risk (concentrated), or add a dedicated AI risk category (AI-specific). The point is not which one you pick, but that the choice is deliberate, approved, and leaves no gaps between risk types or parts of the business.

Know where your AI is. You cannot govern, or scale, what you cannot see. An accurate, firm-wide inventory of AI uses, including low-risk and bought-in ones, is what lets risk be aggregated and the portfolio read. It is a required outcome of the AIRG, not optional housekeeping.

Read the portfolio, do not just receive it. Senior Management should get a small set of AI key risk indicators with named owners, such as uses missing from the inventory, systems running without approval, and drift or incident measures, and an executive view of the portfolio against the risk appetite. The value of a dashboard is in the challenge it provokes, not the comfort it gives.

What good looks like, and what to ask. A stated AI risk appetite, a deliberate taxonomy choice, a complete inventory, and a portfolio view the Board actually interrogates. Ask: where is the appetite breached today, what is in the firm that is not in the inventory, and which indicator would tell us first if something went wrong.

In an ebook: AI Risk Management for Directors.

Decision 4 of 6

Where do people stay in the loop, and the skills to do it

Human oversight matched to risk, and the fluency to exercise it.

As AI does more of the work, where do people have to stay in the loop, and does the firm have the skills to be there?

Match oversight to the stakes. Human oversight is not all-or-nothing. It scales with how material the use is: light, sample-based checks for a minor use; review of every output, or a person in the decision, for a material one. Design the oversight into the use from the start, sized to its risk tier, rather than bolting a reviewer on at the end.

Oversight is only real if the person can act. A human in the loop who cannot realistically question or stop the AI is oversight on paper. The reviewer needs the time, the information, and the authority to say no, and the system has to make its output challengeable, not just presentable.

Build fluency by role, not a single course. MindForge sets skills against the five roles: Executives, Builders, Custodians, Use Case Owners and Business Users each need a different depth. The Board and Senior Management need enough to challenge what they are told, not to build models. A baseline of AI literacy across the firm, and deeper skills where people govern or build, is what makes oversight work in practice.

Culture carries it. People should be able to raise a concern about an AI use without it being career-limiting, and should know who owns what. Scaling AI safely is as much about culture and clear accountability as about controls.

What good looks like, and what to ask. Oversight sized to materiality, reviewers who can actually act, and role-based fluency with a firm-wide baseline. Ask: for our most material AI use, who reviews it, can they really say no, and do they have the skill to know when to.

In an ebook: AI Risk Management for Directors.

Decision 5 of 6

Can you govern an agent that acts on its own

Design-time versus runtime: mandates, institution-owned controls, dispositions and audit.

An AI agent can take actions on its own, not just suggest them. What has to be true before you let it?

You cannot sign it off once. A model that only produces an answer can be reviewed before it ships. An agent acts, across steps, and the risk is in the path it takes, not a single output. So governance has to run while it acts, not only before it launches. This is the line between design-time governance, what you check before go-live, and runtime governance, what holds while it is working.

Give it a mandate, and own the controls yourself. Before an agent acts, it needs an explicit mandate: what it is allowed to do, within what limits, and what evidence it must leave. The controls on it should be the institution's, not the vendor's or the model's own promises, so that the firm, not the tool, decides what is permitted. The most reliable control is to make a forbidden action impossible, not to ask the model to avoid it.

Decide what happens at each action. For a given action the agent proposes, the firm's controls decide its disposition: let it execute, send it for a person to approve, or reject it. High-impact actions sit behind approval; some are never allowed at all. And every action and decision is written to a tamper-evident audit log, so the path can be reconstructed after the fact and someone can be held to account.

Keep a hand on the brake. Oversight and key risk indicators scaled to what the agent can affect, such as actions outside scope, failed tool calls and escalations; a kill switch that someone owns and has tested; and containment that tightens with materiality. Not every agent needs the full treatment, but one that can move money or commit the firm does.

A note on the state of the art. This is live work. I helped a little with MAS's SAFR, a public industry reference for safeguarding agentic finance at runtime, and it relates to a paper I co-authored on runtime governance. The ideas above, mandates, institution-owned controls, action dispositions and tamper-evident logs, are where responsible practice is heading; the ground is still moving, so govern for the direction, not a fixed checklist.

What good looks like, and what to ask. Agents with explicit mandates, controls the firm owns, high-impact actions behind approval, a tested kill switch, and logs you could hand to a regulator. Ask: what can this agent do without a human, what stops it doing what it must never do, and could we reconstruct exactly what it did last week.

In an ebook: AI Risk Management for Agents.

Want to go deeper?

IBF-accredited courses. The deep-dive and technical courses I designed and teach are IBF-accredited and listed on the IBF Programme Finder, delivered through the Association of Banks in Singapore and the Singapore College of Insurance: Banks (deep dive), Banks (technical) and Insurers (deep dive).

Courses I run directly. I also run sessions in-house for a single firm now, and maybe as open cohorts one day, including briefings for leaders and boards, an oversight course for the second and third line, and a technical hands-on day. See the courses.

Adopting generative AI and agentic AI, responsibly and at scale