Under the engagements sits one machine: a typed-graph world-model of how surplus electricity becomes physical value, a staged pipeline that carries a jurisdiction from raw physics to a deployable instrument, and an honesty discipline that keeps every figure as trustworthy as its weakest input. This page is a method explainer — no client work, no returns.
Inari does not start from a spreadsheet of countries. It starts from a typed directed graph: a stack of layers (geology → resource stocks → transformation → value), a small vocabulary of resource types, and a set of enablement edges that say how one resource unlocks another. A library of roughly 20 production chains consumes those resources to produce value, and every jurisdiction simply selects a subgraph — the chains it can realistically run and the resources it has.
That is why one vocabulary describes a 260 km² coral island and a continental federation: only the selected subgraph and the numbers change. And it carries a single organising claim in its edges — cheap surplus electricity is the only resource with outbound edges to several others. Relax the energy constraint and water, soil, protein and refined materials cascade behind it.
Fixed geography — the land and sea a jurisdiction is given. Non-regenerating.
Flows that can be grown or depleted. Energy sits here — with renewables it regenerates, which is what makes it the lever.
The chains operate here — each a way to spend energy (and sometimes other resources) to make something valuable.
Captured economic and sovereign value — the result of the lifting below it.
Each chain ties a piece of engineering physics — the energy needed per unit of output — to its economics and its dispatch flexibility (how easily it can shift power use to match available surplus), so it can be sized against a jurisdiction's surplus and ranked by value density.
Every place is scored on the same ten dimensions — renewable endowment, energy-security urgency, institutional quality, transition momentum, export orientation, water–energy tension, and more. One typed structure is what makes an atoll, an island, and a continental grid comparable.
Energy is upstream of everything that can be grown or refined. The foundation — 100% renewables plus storage — gates every downstream edge: nothing runs until surplus electricity is firm (reliably available, not on-and-off) and abundant. Relax that one constraint and water, soil, protein and refined materials cascade behind it.
A jurisdiction runs through a staged pipeline — one left-to-right spine. Each stage produces a named output that feeds the one after it, so every claim downstream is traceable to a physical or legal input upstream. The legal stage is the gate: structure is a rules problem before it is an engineering one. The financing and value-assignment stages are held back on public surfaces — shown directionally here, detailed only in engagement.
surplus → techno-economics → confidence bands → legal (gate) → finance (held) → rank → value (held) → instrument
The same spine reads as a dependency topology (a map of what feeds into what) — five layers, each one built on the canon (the shared, agreed source data) beneath it. A read-only foundation of ontology and assumptions feeds the surplus model; the surplus model feeds the analysis; the analysis feeds the domain-specific chains; and everything resolves through a single integration layer. Build order runs top to bottom.
Every quantified field travels with a provenance envelope — value, unit, fidelity tier, confidence, evidence, source — so the model's claims are auditable rather than asserted. Where an input is missing, it is shown as an explicit gap and the dependent claim is down-ranked to option-value, never fabricated.
Two rules keep the engine honest. Every figure carries a fidelity tier stating how trustworthy its provenance is — and a composed result is only ever as trustworthy as its weakest input. Most figures Inari publishes are screening-to-feasibility grade (early estimates good enough to compare options, not yet loan-ready); bankable grade (precise enough for a lender to act on) exists, but it is shared under NDA, not published. The gating is on publication, not on the engine's reach.
Indexed to the achievable surplus (= 100), the chart reads left to right: what is recoverable, what survives conversion, the uplift from coordinating it, and the value that stays in the jurisdiction. Bands and relative deltas only — the engine never publishes absolute magnitudes for large interconnected grids.
Indicative figures for triage and first contact — enough to rank options and classify a jurisdiction, not to bank a project.
Backed by real sources and calibrated to the jurisdiction — the feasibility-grade readout that survives a board, IC, cabinet or coalition partner.
The grade a transaction is built on. The engine runs through to it — but it is shared only under NDA, once calibrated for a specific deal.
The second rule is F1, a hard magnitude discipline. Isolated island micro-grids may state absolute surplus and curtailment (surplus power that has to be thrown away) figures. Continental grids and nations are treated relative-only — value is expressed as a delta versus an idealised, frictionless "copper-plate" baseline (same assets, better rules), and the absolute-magnitude slot is suppressed with a note rather than asserted. Together the two rules guard both ends: the fidelity ladder guards inputs against inflated confidence; F1 guards outputs against over-claiming on big interconnected systems.
A scoping conversation is the right first step. An hour to understand the question, then a screening classification — or an honest signal that the framework isn't a fit yet. We will not ship a generic capability deck.
A scoping call is free. Published figures are screening-to-feasibility grade — never investment, financial, or legal advice.