Board education · Facilitator guide · 45–60 minutes

Reading the AI bubble.

A board-level walk through the question everyone is really asking — "is AI a bubble?" — and the governance questions that matter far more than the stock chart. Read it straight through as a board, or switch on the facilitator track to run it as a session.

The premise

Understanding precedes strategy.

You cannot govern or set strategy around a technology you do not understand — just as you would never hire someone with no insurance experience to build an insurance company's strategy. The purpose of this session is to build enough shared understanding that the board can weigh AI risk intelligently and read the news with context.

The risks of AI are real. So is the upside. This is neither doom-saying nor cheerleading — it is the grounding a board needs to tell the difference.

Clean reading view for board members. Switch to Facilitator view to reveal timing, talking points, and discussion questions for each section.

01 · Market framing

The "AI bubble" — where every mind goes first.

Ask "is AI a bubble?" and most people picture the stock market. That is a fair place to start — and the wrong place to stop.

A handful of AI-driven names now make up an outsized share of the major indices. That concentration means broad-market exposure is, quietly, a large bet on a few AI stories playing out. It is worth acknowledging plainly.

But market exposure is a portfolio question for the investment committee — it is not the governance question for this board. Whether AI equities are over- or under-priced tells us almost nothing about whether our use of AI is sound. So we touch the market, then pivot to the risks we actually control.

  • Index concentration makes "the market" partly an AI bet, whether or not anyone chose it.
  • Valuation risk is real but sits with the investment committee, not operational governance.
  • Our exposure that matters is operational: how we adopt, depend on, and govern AI.
A few AI-driven names everything else
IllustrativeNot to scale. The shape of the point — concentration — matters more than any single figure, which shifts week to week.
Facilitator track≈ 5 min

Talking points

  • Open by naming the elephant: yes, the market looks frothy. Give it two minutes, then signal the pivot explicitly.
  • Keep the figure illustrative on purpose — resist debating a specific percentage; the point is concentration, not a number.
  • Land the transition: "the governance risk isn't the S&P — it's how we adopt this inside our walls."

Ask the room

  • When we hear "AI bubble," which of us means the stock market, and which means our own projects?
  • Where does our organisation carry AI risk that has nothing to do with the market?
02 · The dot-com parallel

Real value, real correction.

The dot-com bubble was not a fraud. Real value was created; the internet reshaped every industry. The bubble still burst — and both things are true at once.

The useful question is not "is AI fake?" It plainly is not. The useful questions are who survives the correction, and what the landscape looks like afterward.

Expect the familiar arc: a boom of AI vendors and point solutions, oversupply, then consolidation and failures — followed by a smaller set of durable winners. The technology keeps compounding while the vendor map is redrawn. Planning for the consolidation, rather than betting it away, is the governance posture.

  • Genuine value and a painful correction are not contradictory — dot-com proved it.
  • A vendor boom-and-bust is likely; today's logo may not exist in three years.
  • The winners are rarely the loudest — they are the ones others come to depend on.
dot-com boom shakeout consolidation
IllustrativeThe dot-com arc (grey) as a template for the AI cycle (blue). Timing and amplitude are unknown; the shape recurs.
Facilitator track≈ 8 min

Talking points

  • Anchor with a memory the room shares — a dot-com winner and a dot-com casualty. Make "real value AND a crash" concrete.
  • Move the question from "is it real?" to "who survives?" — that reframe is the whole point of the section.
  • Foreshadow section 3: surviving the shakeout is mostly about what you depend on.

Ask the room

  • Which of our current AI vendors would we bet is still standing after a consolidation?
  • If our main AI provider failed or was acquired next year, what breaks?
03 · Vendor risk & lock-in

Are we building, or just renting?

The governance question is not "which vendors do we avoid." It is "what is our dependency structure" — and what we would lose if a vendor disappeared or pivoted.

Drag the slider to move from full-service dependence toward a layered posture. Watch what the organisation actually owns change — and what happens to its exit risk.

Full vendor dependenceLayered ownership
Exit / lock-in riskHigh

Full-service dependence

Speed today, hostage tomorrow.

Everything runs on one vendor's stack. Fast to start, but nothing of lasting value is retained in-house — if they raise prices, pivot, or fail, there is no fallback and no leverage.

Vendor platforms & models Our data & workflows Model routing & swap capability Proprietary capability & know-how
What we ownBrighter layers light up as the posture shifts toward ownership. The vendor layer never disappears — the goal is what sits on top of it.

The recommended posture is a portfolio: vendor partnerships for speed and scale, plus internal capability-building for resilience and optionality — self-hosted infrastructure where it pays off, model routing, and the ability to swap providers. Renting is fine. Renting everything, and retaining nothing, is the risk.

Facilitator track≈ 12 min

Talking points

  • Run the slider live on screen. Let a board member call out where they think the organisation sits today.
  • Distinguish platform-as-a-service with internal building on top vs. full-service dependence where nothing is retained.
  • Reframe "lock-in" from a purchasing detail to a board-level resilience question with a real dollar and continuity cost.

Ask the room

  • Do we own our data, models, and workflows — or are we renting all three?
  • If our primary vendor doubled its price tomorrow, what is our exit strategy — and how long would it take?
  • Where is renting genuinely the right call, and where should we be building?
04 · FOMO as an internal risk

Discipline beats panic.

The most expensive AI risk is often internal: fear of missing out driving tools adopted before governance, pilots that never mature, and budget committed because a competitor moved.

Good governance grants permission to be deliberate. The goal is not to stop experimenting — keep investing, keep innovating, accept small increments of value. The line is between disciplined experimentation and reactive spending.

Place a recent AI decision on the pad: how well did we understand it, and how deliberately did we move?

Confident
but hasty
Disciplined
experimentation
Reactive
spending
Cautious
but uninformed
Deliberate → Understanding →

X-axis: reactive → deliberate. Y-axis: low → high understanding. Click anywhere to place a decision.

FOMO-driven reactive spend disciplined experimentation the same budget, spent two ways
The lineBoth ends spend money and both take risk. Only one compounds into understanding and retained value.
Facilitator track≈ 10 min

Talking points

  • Name FOMO as a real, quantifiable risk — not a personality flaw. It shows up as stranded pilots and shelfware.
  • Make the permission explicit: governance exists to let us be strategic, not to say no to everything.
  • Use the pad on a real decision the board made recently — the honest placement is the conversation.

Ask the room

  • Which of our AI decisions in the last year was pace-driven rather than understanding-driven?
  • What would "disciplined experimentation" look like as a standing expectation for management?
  • Where are we underinvesting out of fear, not discipline?
05 · Literacy before strategy

Strategy without literacy is theater.

You would never hire someone who has never worked in insurance to build your insurance strategy. Yet organisations routinely set AI strategy with nobody at the table who understands AI.

Would you do this?

Hand your insurance strategy to someone who has never priced a policy, seen a claim, or met a regulator.

Then why this?

Set your AI strategy with no one at the table who understands how the models work, where they fail, or what they cost.

"Use AI to automate claims" sounds great — until reality hits: model hallucination, messy claims data, and compliance constraints on black-box decisions. Then the strategy collapses and credibility is burned. Literacy is what stops that. It makes the strategy smarter, because the right questions start to emerge on their own:

Investment decisions Policy & governance Strategy Literacy — the latticework of understanding
FoundationEverything above rests on the base. Literacy is also bubble protection: it separates signal from noise, so the board neither overspends on hype nor freezes in fear.
Facilitator track≈ 10 min

Talking points

  • This is the punchline — land the insurance analogy slowly and let it sit. It reframes the whole series.
  • Give the concrete failure: "automate claims" colliding with hallucination, dirty data, and black-box compliance limits.
  • Position this session as the first course in an ongoing education series, not a one-off.

Ask the room

  • Who at our strategy table actually understands how these models work and fail?
  • What is one AI question we could not confidently answer today — and who should own it?
  • What would it take to make AI literacy a standing capability on this board?
Run this as a session

Take it into the boardroom.

Read straight through, this is a board briefing. Switch on the facilitator track and it becomes a 45–60 minute session with timing, talking points, and discussion questions built in. Pair it with the interactive tools below to turn understanding into governance decisions with owners.