Production capacity that works like infrastructure for the humanoid-robot era: modular, certified, and portable across factories. The full thesis, working tools, and dated predictions, in the open.
Humanoid robotics and physical AI will reduce the labor-cost advantage of low-wage regions. Where things get made will be decided by product design, BOM structure, energy, logistics, tariffs, risk, and proximity to demand. Not by wages. Moduloa Manufacturing is built for that era.
Products are industrialized into production tiers and portable blueprints. A factory OS controls revisions, layout, stations, tools, quality, and capacity. Certified hubs make capacity comparable and routable. Humanoid robots are the flexible execution layer.
Manufacturing is the long-term direction of the whole brand. The first factory is proof-of-work; the scalable asset is the framework: tiers, standards, certification, training, and routing intelligence.
Stage 0, step 1 (doctrine and artifacts) is complete: thesis, tier model, calculator, scorecard, and the contact map are all published. Next is step 2: 10–20 expert interviews to pressure-test the thesis. The thesis is revised only if something turns out to be wrong, not on a fixed schedule, and the revision comes as a dated announcement noting the stage and step it changed at.
A serious software organization, manufacturing veterans from where manufacturing actually lives, capital that understands the shape, and pilot partners with real products. The team behind the vision →
End to end, the model is one flow: a product comes in as structured data and leaves as routable production. The strip below is the whole pitch at a glance. Select a step to go deeper, or run the full journey in the walkthrough underneath.
The six phases mirror the walkthrough below exactly · each opens into its own page
The interactive walkthrough needs JavaScript. The full journey is readable below.
One structured package starts the machine: the build-tree BOM, the documentation each part class requires, and verifiable supplier identities, exactly as the published intake standard defines them. No sales cycle stands between the submission and a number; an incomplete package is rejected with a machine-readable report, and that rejection is the request for precisely what is missing.
The package is checked against the published rules, supplier lines are confirmed at the source through the confirmation protocol, and batch size, recurrence, product stability, quality requirements and portability map the product to a production tier. Those are the same five inputs the public calculator uses today.
What comes back is not a lump sum: every line costed in the open, a confidence number that states how much is attested versus estimated, optimization tiers for cutting cost, and factory options with their price deltas. Computed, not negotiated. No human touched it.
The order binds to a tier and a quality contract, not to a specific factory. Which floor actually runs it is decided later, by routing.
Process knowledge is captured as a versioned blueprint: stations, tooling, parameters, quality gates. At the higher tiers this is the asset that makes production portable: written once, runnable anywhere certified.
Before any physical setup, the factory OS runs the blueprint against a certified hub's reference model, catching collisions, takt-time gaps and missing tooling while they are still cheap to fix.
The framework's core decision: the blueprint goes to whichever certified hub wins on energy price, logistics, tariffs, risk and proximity to demand. Not on wages. This is the thesis in a single step.
Certification makes hubs comparable: a hub's scorecard states what it can run and how well. Acceptance reserves real capacity against the blueprint's requirements.
The OS translates the blueprint into this hub's floor: station assignments, tool lists, schedules, and the data contracts every machine and worker will report against.
The reconfiguration layer at work: humanoids move cells, fixtures and tooling into the blueprint's layout. That is the physical flexibility that lets one floor change shape without a retrofit. A hypothesis, tracked openly.
The first part off the line is measured against the blueprint's quality gates. Pass, and the run continues; fail, and the deviation is logged against the blueprint version, not against someone's memory.
Fixed, enclosed cells do what they are best at: bounded, repeatable tasks under controlled geometry, the reliable core of today's physical AI.
Between the cells, humanoids do the unglamorous work that keeps a line alive: loading material, swapping tools, clearing exceptions. These are the jobs that are pure cost while only humans can do them.
Every unit's history streams into the OS: measurements, station, tool, operator or robot, timestamps. Every operation that closes a sub-assembly mints and marks its identity on the line. Traceability is not an audit exercise, it is the default data exhaust.
Read the marking requirement →
structured loop · telemetry → the blueprint, versioned
Quality results roll up into the hub's living certification score, the same score routing reads next time. Good work literally earns a hub its next order.
structured loop · quality → the hub's certification score
Logistics closes with the same discipline: what shipped, from which line, under which blueprint version, and the landed cost reconciled against the actual freight and duty invoices, not left as a quote-time guess. A chain of custody a regulated customer can act on.
Parts arrive with their history attached, and a code on the product opens the unit record behind the customer's sign-in: every build step, measurement, and deviation, down to part level. The relationship was never with one factory; it was with a certified capacity network.
Everything the run learned (cycle times, deviations, tooling wear) is compiled into the next blueprint version. Process knowledge compounds like software.
The network effect: an improvement proven at one hub ships to every certified hub running that blueprint. One floor's lesson becomes the network's baseline.
structured loop · improvements propagate to every certified hub
The model above is not only described; parts of it are specified, demoed, and priced, in documents published before the first customer exists. Start with how the money works; everything else is one click deep.
Moduloa is a Norwegian build with no home team. The model grows manufacturing capacity wherever demand and economics justify it; it does not exist to move work from one country to another, and the thesis says so in plain words: learn from Asia, not replace it.
A certified hub in Shenzhen and a certified hub in Trondheim join the same network under the same rules: the same certification, the same blueprint discipline, the same routing. Production goes wherever the economics point, in every direction, including toward the strong Asian manufacturers the thesis names as future certified hubs.
Every serious manufacturing nation is running its own modernization push: Norway rebuilding industrial competence, China advancing its intelligent-manufacturing and embodied-AI standards system, Germany deepening its automation base. An open framework that makes factories comparable and blueprints portable is a contribution to all of them at once. Compatibility with those national programs is a feature of the model, not a compromise of it.
Everything here is published: the economics open-book, the standards drafted in public, the predictions dated and scored. That is what makes the model investable from any continent: there is no hidden agenda to underwrite, only a framework to examine. Hubs, suppliers, and investors join on what the documents say, not on where they are from.
The pre-factory period: thesis testing, network building, learning, and artifacts.
Unlocks when Stage 0 is complete. Teased now so the direction is on the record.
Stage 1 is the proof-of-work: the first Moduloa production line, a real product run through the tier model, the factory OS, and certified-hub discipline, with humanoid execution added as it matures. The first factory proves the framework; the framework (tiers, standards, certification, routing) is the scalable asset. Its own steps, tasks, and dates will open here the way Stage 0's do.
The thesis names its own first artifacts. They live here as public tools; every result is a shareable link, and the models are open for correction.
The first monthly recap synthesizes July's twelve field notes: a first half tracing the humanoid execution layer, where the value keeps landing beside the robot, and a second half that turns inward to price the framework itself: capacity, admin, landed cost, and currency. Read it →
The product-lifecycle exploration converged with the flagship: structured production data from first part to last is now the unit record. Per-part identity under the marking requirement, build history minted as sub-assemblies close, and customer APIs. The brand keeps a short list, and lets work converge. The merge note →
A public journal for research on humanoid robotics, manufacturing economics, and reshoring, refreshed each week by an automated deep-research pass and recapped each month. First note: the state of humanoid work in mid-2026. Read the journal →
The honest requirements: a serious software organization with IT/OT security at the core, manufacturing veterans from Asia and beyond, capital that understands platform economics, and pilot partners with real products. Deadlines and weekly automated review added to the predictions register. Read it →
The thesis boiled down to its falsifiable core: claims about humanoids, manufacturing economics, Norway, and Moduloa itself, frozen with audit horizons at 2031, 2036, 2046, and 2056. Read them →
The two artifacts the thesis names for Stage 0, built as client-side tools with URL-encoded state so every result is a permanent, shareable link. The models are deliberately simple drafts, published to be challenged in the expert interviews.
Vision, production tiers, factory OS, certified hubs, business model, risk register, and the Stage 0 execution plan, written down so it can be reviewed honestly later. Read it in full →
Add to this
Corrections, evidence, and disagreement are welcome. This is knowledge in the open. Anyone can read; sign in with GitHub only to post or react, and it appears here instantly.