BlackSwan

Projects

Make the mission fundable before you build it.

A model that turns ‘we will do good’ into a 30-year plan a funder can underwrite, with the failure modes named.

8.5 million meals, modeled over 30 years. What a single nonprofit greenhouse network is projected to deliver in the base case, on a plan stress-tested to stay solvent before a dollar is raised.

Context

The concept is Cultivate the Future: a nonprofit network of automated greenhouses that grows food for the community and trains people to apply AI in the real world, run on an open-source agentic stack that handles the compliance and back-office paperwork most nonprofits drown in. The real question behind it was never ‘is this a nice idea.’ It was whether the thing can survive thirty years, fund itself, and how much good it actually buys per dollar. A funder, a board, and a partner all need that answered before they commit, and most mission pitches cannot answer it.

The problem

Mission-driven plans usually arrive as a vision and a wish. The hard parts, whether the grants actually come, what happens in a bad year, what each unit of impact really costs, get left to faith. Good intentions are not a plan anyone can underwrite, and the gap between the two is where ambitious nonprofits quietly fail.

What I built

The same kind of stress-test I bring to a power plant or a product, pointed at a mission. The model runs the full 30-year build-out, year by year and site by site: what it costs to build and operate, where the money comes from, and which shocks would sink it. It compares the venture run lean on an automated compliance stack against the traditional staff-run version, so the automation thesis is tested, not assumed. That stack turned out to be the single lever that decided whether the venture survived, and it became its own piece of work: see NPOS. And the model is honest by construction. Revenue that is not yet contracted is never counted, and carbon the organization sells to someone else is never claimed as its own.

Results

The plan holds. In the base case the network delivers 8.5 million meals over 30 years at about $2.50 in net cost per meal, plus roughly 55,000 tonnes of net carbon it can actually claim as its own. It stays solvent across the market scenarios it was tested against, with the rare severe shock flagged as the edge to watch. And the riskiest piece of the concept, the earned-revenue tech platform, turns out to be pure upside: the organization is viable without it ever working. The part most people would have bet the venture on is the part it does not need.

What this shows about working with me

Maps to: Delivery rescue

This is the same first-principles move whatever the domain. Take a big, ambiguous, expensive question, find what actually decides the outcome, and turn it into numbers someone can commit on before the money moves. Here the goal was meals, not margin, and the engine was the automation thesis at the center of the practice. The muscle is the same one that rescues a stalled build.

Book a free diagnostic. It starts with the same question: what would it take to make your plan one a serious person could underwrite?

Get a free diagnostic