Fix the Roof on a Sunny Day

A founder asked us: business is good this year, so why not take on more mandates while you can?

Let me state the view up front. A good year is exactly the reason not to take on more. The slack a tailwind leaves in your hands should go into structure, not into the harvest. Why? Because structure can only be worked on in fair weather.

Fixing the roof on a sunny day means using the best years to put up the structure for the next eight. You can patch a leak in the rain, but the movements come out distorted: you mend only the part that hurts most, and nothing can be pulled down or altered. A sunny day is different. You can dismantle at your own pace and build at your own pace. Get it wrong and there is still time.

So what exactly is being worked on? The structure, not the business.

Standing at the 2018-to-2026 mark, we carry two forms. One is the mature form: a boutique investment bank focused on high-barrier hard technology and physical AI, with a steady organisation, a delivery process you can retrace and checkpoints you can reconcile. The other is the evolving form: boutique industrial capital that uses professional human capital to amplify the efficiency of capital allocation. It fuses advisory work with strategic investment and holds quality assets long term with proprietary capital. Each feeds the other.

Put Your Own Money In and You Look at a Deal Differently

Put proprietary capital into the companies we serve as co-investment, and a wrong judgement lands on our own book first. In other words, a rule is more reliable than self-discipline.

This can only be done in fair weather.

Fast in Delivery, Slow in Choosing

The tempo of the second eight years is one fast, one slow.

Fast in delivery. The same milestones, reached in a shorter cycle and with higher certainty, one quarter doing the work of a year — which gives a year three more chances to correct course. Why the hurry? Because the speed at which tools spread has changed by an order of magnitude. The last generation of production tools took decades to cover the market. This generation, once a good thing is accepted, spreads in months: the channels are already there, and the users were trained by the generation before. When things spread that fast, the work has to keep up.

Slow in choosing. A general on the march does not chase rabbits. We do twenty to thirty mandates a year, half of them new and half of them existing clients coming back for the next round. That half — the repeat mandates — is our report card. Our headcount can grow. The number of projects cannot. Add projects and the attention each one gets is thinned out. Restraint is the first word we have written for our second eight years.

Capability Has to Be Built into the Organisation

When the tools of production improve, the relations of production follow. Every generation of new tools brings a new generation of organisational method: lean production, Six Sigma, OKRs, Rendanheyi, the flywheel — they all came this way. We are working out our own answer for an AI-collaborative organisation: the Glacier Model (Orchestrator Model, still being defined).

Its plainest component is the subtract-only list (Stop Doing List): whatever should no longer be done, written down as a prohibition and reread by everyone at every review. Methods can iterate; boundaries only grow.

The best young people are the frame of the organisation, and the organisation is their soil. Colleagues one or two years in can follow three to five projects through a full cycle because eight years of method, sequence and standards have settled into the organisation. In a tailwind these beams look the least urgent. But a tailwind is also exactly when they are cheapest.

The right use of a good year is to invest in future structure, not to consume the present tailwind. Fix the roof on a sunny day. Do the arithmetic over ten years.