The Board's Most Undervalued Asset

The path to the boardroom

If your plan for a first board seat is to wait to be asked, the data says you will be waiting a long while.

Egon Zehnder's guide on first-time boards puts the average search for a first corporate seat at more than two years. Turnover is slow, the competition is senior, and almost nothing about the process is advertised.

Two things move it faster. First, boards are no longer only buying former CEOs. They recruit for specific gaps: digital, AI, regulatory, scientific, so a specialist with one deep edge is often more appointable than a generalist with a big title. Second, most seats travel through networks and search firms rather than applications, so the people who get the call are the ones who have told their network, their mentors and the search firms plainly that they want to serve.

Read the piece →

People committees: boards' most undervalued assets

Boards bring real discipline to finance and risk, and far less to the decisions that shape who leads the company and the culture they inherit. A new Egon Zehnder article in HBR shows why succession, leadership and culture so often escape the rigour they deserve, and what the boards that get it right do differently.

Five shifts:

1. Clarify the mandate. Succession, compensation and culture demand different thinking; don't bundle them into one vague committee.
2. Create real conditions for independence. Closed sessions without management, and recusal when the discussion is about the people in the room.
3. Make succession continuous, not episodic. A year-round cadence and real exposure to the pipeline, not one annual review.
4. Widen the field. Don't let the chair's network stand in for a proper search.
5. Build the right capability mix. Put leadership, culture and compensation expertise on the committee itself.

The standard boards bring to the balance sheet is the one they can bring to who runs the company next.

Read the piece →

The switch to graph engineering in AI agents

Peter Steinberger's (creator of OpenClaw) viral post sparked a wider essay on why AI agent architecture is shifting from single self-improvement loops to networks of loops, and what that shift does and does not fix.

A single tweet captured a field's shift. Steinberger's post, "Are we still talking loops or did we shift to graphs yet?", gathered thousands of likes, capturing agent builders' recognised shift from single loops to networks of loops ("graphs").

Four failure modes of a single loop. The essay names four structural failures: Goodhart's law (an optimised metric stops measuring what it should), blindness to whether the target is correct, conflict between independent loops, and undetected measurement decay.

For boards and leaders. For investors backing agent infrastructure and AI-ops tooling, this reframes what a defensible "self-improving" system needs: a governed network of loops with independent audits. The harder point, that even a good graph fails without ground-truth anchors, is a useful diligence question for any autonomous self-improvement pitch.

Microsoft's AI transformation playbook

Microsoft published a 44-page playbook on becoming a "frontier firm", its methodology for making AI transformation stick. It rests on five elements:

• Start with strategy, not AI. Point AI at the business goals you already have, and judge it on the real value it creates, not on how much activity it generates.
• Get it out of pilots. Build a way of working that spreads AI across the company through people and systems, so it delivers at scale instead of stalling as a handful of experiments that never ship.
• Bring your people with you. Reshape roles, how managers lead, and how people learn, so AI expands what your team can do rather than hollowing it out.
• Turn your edge into an asset. Capture what genuinely sets your company apart, and build it into how you test and improve AI, so that advantage compounds instead of leaking away.
• Keep every action reversible. Put guardrails on each AI agent so every action it takes can be traced, checked and undone, before you scale it up.

For boards and leaders. This is hard. It asks people to fundamentally rethink how they work, and reaching the frontier takes courage and sustained leadership. But when leaders give people clear intent and real room to act, they rise to it, and reinvent how the work gets done.

Read the piece →

A FINRA for AI

Demis Hassabis, CEO of Google DeepMind, has proposed the creation of an independent regulatory body to oversee the release of advanced AI models, modelled after the Financial Industry Regulatory Authority (FINRA). It would assess risks and develop best practices for AI deployment, initially letting labs voluntarily submit their models for review before release, with expectations of a framework in place by year-end.

Read the piece →