Roadmap

What's proven, what's next, and where this is headed.

Inari is built one jurisdiction at a time, and the method is built one capability at a time. This is an honest read of where the programme stands today — the engine can now be tested, not just trusted, with New Zealand as the first validated jurisdiction — and the ambition stacked behind it. We hold to now / next / later rather than hard dates: the sequence is committed, the calendar isn't.

Now / Proven
Tested, not
just trusted —
and falsifiable.
Working today

The engine can now be tested, not just trusted. First validated jurisdiction: New Zealand.

Most models ask you to trust them. This one can be tested. The method that proved out end-to-end on a Pacific micro-island has now cleared a full validation pass on New Zealand — the first jurisdiction to earn a validated mark. Every headline now carries a validation manifest: sensitivity checks, pattern tests, a remove-the-gate falsification (a deliberate test that tries to break the claim by removing a key assumption), and stated uncertainty bands. The engine has also grown wider — it now reasons about water, industrial demand, resource endowment and fiscal resistance — and reads the governance friction that shapes what's actually buildable. This is the credibility upgrade: the analysis stands up to being challenged, not just presented.

IFalsifiable, not just presented

Every headline carries its own stress test.

The differentiator: our figures can be attacked. Each headline ships with a validation manifest — a sensitivity check on the assumptions that move it, pattern tests against known-good behaviour, a remove-the-gate falsification that shows what breaks the claim, and honest uncertainty bands. New Zealand is the first jurisdiction to clear that pass end-to-end and earn a validated mark. If a result can't survive being challenged, it doesn't ship as a headline.

Status: Validated · New Zealand (first)
IIThe method, end-to-end

From data to a defensible answer.

The Surplus-to-Structure chain (the method for turning surplus power into durable industrial assets) — dossier → analysis → populated deliverable — has been walked all the way through, first on a live Pacific micro-island and now on New Zealand. The honesty discipline travels with it: figures render as screening-grade (early-stage estimates, not final numbers), gaps show as gaps, and the case is made on value kept home and resilience, never a headline return. The shared ontology underneath is jurisdiction-agnostic — roughly two-thirds of the framework carries to any new country unchanged.

Status: Proven · micro-island & national scale
IIIA wider engine that reads friction

More of the picture, measured.

The analytical core is a mature, continuously tested engine — and it now sees more. It reasons about water security where it's the binding constraint, the pull of real industrial demand on how surplus gets sized, and a jurisdiction's underlying resource endowment. On top of that sits a governance-friction layer that reads how incumbent resistance and political drag shape what can actually be built. Wider coverage, same honesty discipline, all under the validation harness above.

Status: Built · water · demand · endowment · friction
Next / In flight
More places
validated. Make
it answerable.
In progress & near-term

Validate more of the map. Make it answerable.

With the first jurisdiction validated and the engine widened, the next moves are breadth and access: run more jurisdictions through the same validation pass, and put a grounded assistant in front of the method so people can ask it questions directly. Each step is additive — it doesn't disturb what's already proven.

APipeline breadth

From one validated case to a cohort.

New Zealand is validated; the pipeline now carries the same pass to a wider Pacific micro-island cohort and on to Australia as the first industrial-nation case — anchored on rare-earth and critical-minerals processing (large, patient buyers that soak up surplus power on long contracts). Each jurisdiction re-instantiates the method and has to earn its own validated mark; the diversity of these cases is what stress-tests the framework's range.

Pipeline: NZ validated · Pacific cohort · Australia (rare-earth)
BAsk the engine

A grounded assistant for the method.

A public assistant that answers questions about the Inari method and the public Pacific work, and points genuine interest toward a scoping conversation. It speaks only from a curated, publishable knowledge base — about how the method works, never about client specifics or returns — so it stays useful without overclaiming.

Scope: Grounded Q&A · method only
Later / Ambition
Direction of
travel — not
a commitment.
Ambition, not a date

Deeper coverage, live answers, and Inari as a tool others can call.

Further out, the direction is clear even where the timing isn't. We want the engine's reasoning to deepen across the whole validated pipeline, the assistant to answer from live model outputs rather than a static brief, and the publishable layer to be callable by other systems. These are stated as ambition, deliberately — the order is set, the dates are not promised.

iDeeper engine coverage

The full pipeline, at depth.

Carry the widened engine and its validation discipline across every jurisdiction in the pipeline, and push the reasoning deeper — richer treatment of the constraints and friction that decide what's actually buildable, each new layer held to the same falsifiability test. Credibility before numbers, everywhere.

Direction: Depth & validation across the pipeline
iiLive engine-query assistant

Answers from the live model.

Graduate the assistant from talking about the method to surfacing real, published figures on request — rendering a single public-safe metric as an evidence card, drawn only from a curated, publishable view. Built only once visitor demand for real numbers is demonstrated, and behind the same confidentiality wall.

Direction: Evidence cards from publishable data
iiiInari as a callable tool

A data source in the agentic ecosystem.

Expose the same publishable layer so other systems and AI agents can call Inari as a tool — a partner's assistant or a DFI (development finance institution) analyst's agent querying the public-safe slice directly. A distribution play, behind the identical wall: the publishable view is callable; the bankable engine (built to financing-ready standard) stays private.

Direction: Open the public layer (MCP)
A note on candour
What's public.
What needs
an engagement.
Screening and feasibility are public. Bankable work is under engagement.

Screening-grade analysis and the public showcase work are open — that is what this site, the cases, and the assistant share. Bankable figures — returns, deal structures, the value-side detail — render only under engagement, at calibrated fidelity, under NDA. This isn't coyness; it's the same honesty discipline the method runs on. We would rather show you exactly how far the public layer goes than imply it goes further.

Engage

Want to see where your jurisdiction sits on this map?

A scoping conversation is the right first step. An hour to understand the question, then a proposed product — or an honest signal that the framework isn't a fit yet. No generic capability deck.

Request a scoping conversation → Browse products

A scoping call is free. The first deliverable is fixed-fee.