Capital Allocation is Your Strategy

1. Every project looks profitable. That's the problem.

A group investment leader at one of our longtime customers described the pattern plainly. Inside a large project team, hundreds of small assumptions each drift a little optimistic, because everyone wants the project to happen. No single assumption is wrong. The sum is. And when every request comes with a strong business case, the challenge isn't finding good investments. It's knowing which ones deserve capital most when the CFO caps the total.

That's a portfolio question. Spreadsheets and approval workflows were never built to answer it.

2. The risk is usually in the risk assessment.

Another customer walked me through a major new production line that came online almost exactly as new tariffs hit its main export market. The Monte Carlo simulation had a P10 outcome. Nobody had asked whether that outcome was a scenario they actually believed would happen, or what they would do if it did.

Distributions aren't scenarios. Averages aren't marginal economics. If the tool doesn't help decision-makers live through a plausible bad case before they commit 25 years of capital, it hasn't done its job.

3. The bottleneck is data, not technology.

When we asked customers where their trust boundary sits with AI-supported decisions, the answer was unanimous and immediate: the data. Pulling operational data for a single site can take days. Base cases are one to two months old by the time they're used. Interoperability between systems is worse than any vendor assumes.

I take that as a design brief, not an excuse.

Our vision hasn't changed: to be the Capex Strategy Execution Platform for the entire capex cycle, from strategy to budgeting to management, and the decision hub in the enterprise's increasingly automated ecosystem.

Here's how I read that vision after this summer.

Capex strategy is the framework everything else optimizes within. ERP, asset management, supply chain optimization: these systems will increasingly run themselves. But automation can optimize inside a frame; it can't set the frame. Someone still has to decide which businesses to be in and which assets to own ten years from now. In a capital-intensive company, that decision is the capex strategy. That's the job we anchor.

AI does the heavy lifting. Humans keep the decision. Our customers were clear: recommend, never decide. I agree, and not as a concession. In hotel pricing, an algorithm can reprice a room a hundred times a night and be wrong occasionally at little cost. A capex decision is the opposite. Capex decisions can't be undone, there's no single right answer, the bet changes the world around it, and someone has to answer to the board. What AI should do is what our customers asked for: build base cases in hours instead of weeks, generate and challenge alternatives, rank them per dollar of capital under a real constraint, run the cross-site comparisons that today never happen, and act as the neutral voice that questions the optimistic drift inside a project team.

Connect, don't replace. Our customers don't want another system of record to migrate to. They want a constantly connected picture of assets, actuals, prices, and project status, not a one-time load that's stale three months later. So we're building Weissr to plug into the systems they already run.

Replace the spreadsheet, not the workaround. For most of these companies, the base case for a capital plan still lives in Excel, rebuilt by hand, disconnected from the actuals and assets it's supposed to represent. That's not a missing feature. It's the absence of a system, and it's the starting point we design against.

Reach the market, not just the mill. The most consistent request I heard was for external intelligence: supply and demand, competitor moves, trade flows, footprint. Decision-makers don't trust a model on its own. They want a strategic narrative: What's the trend? What's the market? Why does this improve our position? A decision hub that stops at the fence line of the plant hasn't earned the name.