How to Brief Your Board on AI

I've sat through dozens of board AI briefings. Most of them fail in the first two minutes.

The executive opens with a technology explanation. Large language models. Neural networks. Fine-tuning parameters. The board's eyes glaze over. By the time you get to the business case, you've lost the room.

The Five Principles of Strategic AI Briefing fix this.

Principle one: lead with commercial consequence. Open with the dollar figure. "This initiative targets $2.4M in annual margin improvement" gets attention. "We're deploying a transformer-based NLP model" doesn't.

Principle two: frame risk alongside opportunity. Pull the relevant entries from your Board AI Risk Register and present them together. Directors approve initiatives they can assess. If you only show upside, they'll assume you haven't done the work.

Principle three: benchmark against competitors. Show where peers and competitors are deploying AI in this same area. Nothing motivates a board like competitive displacement.

Principle four: define success and failure criteria upfront. 75% of AI pilots never reach production. Your board knows this stat. Address it directly. What are the metrics that prove this works? What are the kill criteria if it doesn't? What's the maximum capital at risk before a go/no-go decision?

Principle five: present the ask with options. Don't present a single path. Give the board two or three options with different risk and investment profiles. Let them choose. That's governance in action.

Keep the briefing to 15 minutes of presentation, 15 minutes of discussion. If you can't explain it in that window, you haven't distilled it enough.

The Board AI Risk Register does double duty here. It gives directors a familiar framework to assess AI proposals against. They've seen risk registers for cyber, for compliance, for operational hazards. AI risk presented in the same format removes the mystique and enables informed decision-making. Each well-structured briefing compounds board AI literacy over time.

Board alignment determines whether AI initiatives get funded, scaled, or killed. A poor briefing doesn't just waste meeting time. It can set your AI programme back by quarters.

Directors have a fiduciary duty that now extends to AI oversight. They need sufficient information to discharge that duty. When you brief well, you're not just seeking approval. You're giving directors what they need to fulfil their legal obligations.

The consequences of poor briefings compound. Boards that don't understand AI tend to either over-restrict (killing every proposal out of caution) or under-govern (approving everything without scrutiny). Both outcomes damage the organisation.

53.4% of C-suite executives hide their AI use. If your board briefing doesn't acknowledge shadow AI and present a governed alternative, you're ignoring the elephant in the room.

From a competitive standpoint, boards that receive effective AI briefings make faster, better-informed investment decisions. While competitors are stuck in approval loops because their boards don't understand the opportunity, your organisation is deploying.

The ASIC governance expectations add urgency. If a regulator asks your board about AI oversight, the quality of your briefing materials becomes evidence of governance capability or governance failure. Good briefings don't just inform a single decision. They build a board's AI literacy over time, so each subsequent proposal moves faster and with sharper questions.

A CFO at a $300M logistics company restructured her AI briefing using the Five Principles. She opened with "$1.8M annual savings from automated freight routing" instead of explaining the optimisation algorithm. She mapped three risks from the Board AI Risk Register, showed competitor adoption rates, defined a six-month pilot with monthly metrics, and presented three investment tiers ($150K, $350K, $600K). The board approved the middle tier in 12 minutes and asked for quarterly updates.

A CIO briefed his board on generative AI adoption by leading with the shadow AI problem: "42% of our staff are already using ChatGPT with company data and no controls." He presented the risk, the governed alternative, and the cost of doing nothing. The board approved the enterprise AI platform that same meeting.

A CEO's briefing failed when she presented an AI customer analytics tool with a 24-slide technical walkthrough. The board requested a "plain English" version and deferred the decision by three months. The delay cost the organisation first-mover advantage in their market segment.

One board I advise now uses a standard AI briefing template requiring: commercial impact (paragraph one), risk register entries (paragraph two), competitive context (paragraph three), success and kill criteria (paragraph four), and the investment ask with options (paragraph five). Every AI proposal follows this format. Decisions happen in the meeting, not after it.