The company sequence

Software first. Then steel. Then ships.
The sequence to an American shipbuilder.

What must stand, and in what order, before Veldarium builds a ship?

If Veldarium is funded, the first money fills four seats in-house (industrial software; data and integration; configuration management; production control and planning) and opens one design partner’s work area, where governed build state must survive the partner’s own release authority on real records. Everything after waits on evidence from the step before: reasoning over that state, a whole program, one production cell in leased industrial space, a production unit of several cells, the search for land and a first yard, and only then vessels.

Work exists on 3 of 7 rungs, all of it software on synthetic records. No rung carries a date.

Rung 0 · Governed build state to Vessels

Next: Real records, live decision

The 7 rungs

Each rung opens on evidence from the rung beneath, never on capital or a date.

Rung 0
Governed build state
SyntheticCapital phase: Software foundationGate: proof level 4, real records, live decision
For every work package in one area: the governing revision, whether material and evidence are accepted, whether work can start and who may release it, read from the systems that own those facts.
Stands onThe records a yard’s own engineering, material, production and quality systems already hold.
Exists
Tested software for readiness, change impact, evidence and authority
A synthetic vessel and build hierarchy to run it on
Three answers only: ready, not ready, or cannot establish
Does not exist
A run on a customer’s real records
A connector to any yard system
The standalone application
People
Software engineers · A design partner’s production or planning lead · Configuration and data specialists
Capital
Engineering salaries · Hosting · Security review · Legal and data rights
Equipment
Compute and secure hosting · No production equipment
Measured by
Time from a change to a defensible readiness answer · Answers that survive the owner’s review · Released work later found blocked
How it fails
The model cannot represent a real yard’s records · Its answers do not survive the yard’s own review · It becomes another report nobody acts on
Opens the next rung whenOne real release decision on a design partner’s work area, rebuilt from its own records: every fact traced to its owning system, a named owner, and failure criteria written before any record is read.
Rung 0: why here, the gate test, partners and risk
Why here
Everything above acts on this state: reasoning reads it, a cell is released from it, a yard runs on it. It is also the cheapest rung to disprove: one work area, read-only, no steel.
How the gate is judged
Before any record is read, the partner and Veldarium write down the decision, the baseline, what passes and when to stop. It passes only if the partner’s release authority can trace every fact in the answer to the system that owns it, and nothing that should have been refused was answered.
Leaves for the rungs above, once its gate opens
A governed record of build state · Connectors to the systems that own the facts · Readiness answers that survived review · A decision record with named owners
Partners it needs, by kind
A shipyard or fabricator willing to open one work area’s records · An independent security reviewer · Legal counsel for data rights and handling
Main risk
Its answers do not survive the yard’s own review. The gate is one decision on one work area, so a failure shows early and costs little. Every answer runs beside current practice, checked by the people who rebuild it by hand today, and stopping is a written outcome, not a crisis.
Rung 1
Operational reasoning
TargetCapital phase: Software foundationGate: proof level 4, real records, live decision
A model working over governed state: gathering what a decision needs, naming conflicts and unknowns, proposing one option and what would overturn it. It never holds release authority.
Stands on
- A governed record of build state rung 0
- A decision record with named owners rung 0
Exists
A deterministic, synthetic reasoning demonstration, no model connected
Published authority boundaries and the tests that hold them
Does not exist
A connected model
Reasoning over live state
A real decision supported from question to recorded outcome
People
Software and machine-learning engineers · Reviewers from engineering, production and quality · An independent evaluator
Capital
Engineering salaries · Model and compute · Independent evaluation
Equipment
Model access and evaluation compute
Measured by
Proposals checkable from their cited evidence · Refusals that were right · Decisions where authority stayed with the named person
How it fails
Reasoning that sounds right and cannot be checked · Authority drifting from the named person to the recommendation · An unknown rounded up to a pass
Opens the next rung whenA real decision-support workflow in which the reasoning, its evidence, its refusals and the human decision can all be inspected, with the person who accepted or rejected it named.
Rung 1: why here, the gate test, partners and risk
Why here
Reasoning is only as good as the records it reads, so it comes after they can be trusted, and never in their place.
How the gate is judged
Every proposal, refusal and human decision on the real workflow is logged. The evaluator checks each proposal against the evidence it cites, and the gate fails on any decision whose authority moved from the named person to the recommendation.
Leaves for the rungs above, once its gate opens
Prepared decisions with cited evidence · Authority boundaries held on real decisions
Partners it needs, by kind
Model providers, reached through a bounded interface · An independent evaluator · The partner’s engineering, production and quality reviewers
Main risk
Authority drifting from the named person to the recommendation. Reasoning has no write path. It proposes into the record and a named person decides, a boundary Veldarium’s tests already hold on synthetic decisions. The rung adds an evaluator from outside the team before any real decision is supported.
Rung 2
Program-wide governed state
TargetCapital phase: First deploymentsGate: proof level 4, real records, live decision
Governed state across the handoffs where programs lose time: engineering release to material, material to work package, work package to inspection and acceptance, for a whole program.
Stands on
- A governed record of build state rung 0
- Connectors to the systems that own the facts rung 0
Exists
A synthetic material-flow model that reads readiness from the kit
A synthetic program view naming each constraint and its owner
Does not exist
Procurement, inventory or production integration
A program running on it
A team using it without Veldarium in the room
People
Integration engineers · Planners, material control and quality leads at the partner · Security and data-rights reviewers
Capital
Implementation and integration engineering · Connector development · Security and support
Equipment
Integration and hosting infrastructure · No production equipment
Measured by
Configuration effort per deployment, falling or rising · Reconciliation with the systems that own each fact · Repeat use by the decision owners
How it fails
Per-customer configuration becomes the business · Integration cannot reach the depth the answers need · Industrial sales cycles outlast the runway
Opens the next rung whenThe layer holds under engineering, material, production and quality authority on a live program, used repeatedly by the people who own those decisions, with a working exit path.
Rung 2: why here, the gate test, partners and risk
Why here
One decision proves the method. A program proves it survives real authority boundaries, real volume and real change.
How the gate is judged
Measured on the live program against baselines set before anything was connected: repeat use by the people who own the decisions, clean reconciliation with every owning system, and an exit path that has been exercised rather than assumed.
Leaves for the rungs above, once its gate opens
Program-wide material and production state · Reusable connectors and identifier mapping · Measured baselines for time, rework and blocked work
Partners it needs, by kind
A design-partner program with planning, material and quality owners · The vendors of the systems that own each fact, for read access
Main risk
Per-customer configuration becomes the business. Configuration effort is measured from the first deployment. Connectors and mapping have to carry from one deployment into the next; if the effort rises instead of falling, the sequence stops at this rung and the record says why.
Rung 3
Bounded physical cell
Long-termCapital phase: Physical R&DGate: proof level 5, physical process
One process (cutting, welding, fit-up or scanning) in leased industrial space, proving the whole loop on steel: governed state, an authorized instruction, physical execution, a returned measurement, inspection and an accepted state.
Stands on
- A governed record of build state rung 0
- Readiness answers that survived review rung 0
- Authority boundaries held on real decisions rung 1
- Measured baselines for time, rework and blocked work rung 2
Exists
Nothing yet.
Does not exist
A facility
Equipment
Qualified operators
A physical result of any kind
People
Manufacturing engineer · Qualified operator · Safety owner · Quality inspector · Software and controls engineer · Maintenance support
Capital
Equipment purchase and installation · Facility fit-out · Safety validation · Process qualification and test material · Manufacturing and controls engineering
Equipment
One machine or robot · Fixtures and tooling · Sensors and metrology · A safety system · Facility power and space
Measured by
Cycle time · First-pass yield · Dimensional outcome · Uptime · Rework · Inspection result · Exception rate
How it fails
The process cannot be qualified · Instructions from the released model do not match what the cell can do · Evidence returns too slowly or too sparsely to govern from
Opens the next rung whenQualified people, an approved instruction, safe-state behavior, physical execution, returned evidence and inspection, with the result recorded whether it passes or fails.
Rung 3: why here, the gate test, partners and risk
Why here
Software evidence has to survive contact with steel before anyone builds more of it. One cell is the smallest physical test that can fail honestly.
How the gate is judged
Judged in the cell, on test articles: the process qualified to its written procedure, every instruction traced to the released state, every measurement returned to the record, and a failed run recorded as plainly as a passed one. The safety owner and the inspector sign, not the team that built the cell.
Leaves for the rungs above, once its gate opens
A qualified process run from governed instructions · Measurements returned into the governing record · Safe-state and inspection procedures that held
Partners it needs, by kind
A machine or robot vendor and an integrator · A process qualification body · A landlord or host facility for the industrial space
Main risk
The process cannot be qualified. The first cell runs one process, chosen because it can be qualified and inspected, on test articles before any production work. A failed qualification costs one cell, and the record shows where it failed.
Rung 4
Integrated production unit
Long-termCapital phase: Physical R&DGate: proof level 6, repeatable physical output
Several processes run as one controlled unit, such as a panel line: instructions from the released model, machine and sensor state read back, as-built geometry checked against design.
Stands on
- A qualified process run from governed instructions rung 3
- Measurements returned into the governing record rung 3
- Safe-state and inspection procedures that held rung 3
- Program-wide material and production state rung 2
- Prepared decisions with cited evidence rung 1
Exists
Nothing yet.
Does not exist
Everything rung 3 needs
Machine programs generated from a released model
Repeatable output
People
Industrial and production engineers · Robotics and controls engineers · Fitters, welders and pipefitters · Production supervisors · Metrology and quality
Capital
Cell purchase and installation · Controls integration · Operator hiring and qualification · Maintenance capacity · Working capital for material
Equipment
Several process cells · Material handling · Line fixtures and tooling · Metrology and scanning · Controls and maintenance
Measured by
Time, cost, quality and rework per unit against a baseline · Line balance · Handoffs where the record fell behind the steel
How it fails
The line cannot be balanced · Automation fits less of the work than planned · Rework erases the time the unit was built to save
Opens the next rung whenRepeatable physical output, with time, cost, quality and rework measured against a baseline set before work began, and an accepted as-built configuration.
Rung 4: why here, the gate test, partners and risk
Why here
A cell proves a process. A unit proves that processes hand off to each other without the record falling behind the steel.
How the gate is judged
Measured unit by unit against the baseline set before the first unit: time, cost, first-pass yield and rework, with the as-built configuration accepted by quality rather than by the people who built it.
Leaves for the rungs above, once its gate opens
Repeatable output measured against its baseline · A qualified production crew · Handoffs that keep the record level with the steel
Partners it needs, by kind
Equipment and controls integrators · Material suppliers · Training providers for the trades
Main risk
Rework erases the time the unit was built to save. Rework is measured from the first unit against a baseline set before work began, so a unit that does not pay shows early. Automation is scoped to the work it has been shown to reach, and people keep the rest.
Rung 5
Integrated shipbuilding capability
Long-termCapital phase: Production capabilityGate: proof level 7, integrated production
A first yard: engineering, supply, production, workforce, facilities and quality working as one institution on one governed state, partnering where others are better and owning only where ownership has been shown to matter.
Stands on
- Repeatable output measured against its baseline rung 4
- A qualified production crew rung 4
- Handoffs that keep the record level with the steel rung 4
- Reusable connectors and identifier mapping rung 2
Exists
Nothing yet.
Does not exist
A yard or facility
A workforce
Supplier agreements
A quality, class and regulatory approval path
People
Naval architects · Systems, structural and electrical engineers · Supplier development · Production control and planning · Skilled trades across every line · Facilities and maintenance · Program leadership
Capital
Facility acquisition or lease · Workforce hiring and training · Supplier financing · Quality systems · Insurance and compliance
Equipment
A production facility with lift and launch capacity · Steel and outfit lines · Cranes and heavy lift · Utilities and power
Measured by
Throughput against plan · Supplier on-time and acceptance rate · Qualified trades available against demand · Rework hours
How it fails
Capital outruns the evidence base · The workforce cannot be recruited and qualified in time · Demand is too unstable to keep capacity loaded · No site clears the physical gates at a cost the program can carry
Opens the next rung whenOne bounded vessel program the evidence supports, on a site that can build it, with demand, a design mature before construction, a qualified workforce, supplier capacity, quality and regulatory authority and working capital each established, not assumed.
Rung 5: why here, the gate test, partners and risk
Why here
Only repeated, measured units justify the land, the workforce and the supplier commitments a shipbuilder carries. A yard chosen before that evidence would be sized to a guess.
How the gate is judged
Each condition is judged by the party that holds it: demand by a committed customer, design maturity by engineering review, the workforce by qualification records, supplier capacity by written commitments, the approval path by class and flag authorities, working capital by the accounts. A site is judged against criteria published before any location is studied. It is out if it fails a hard gate, whatever else it offers: channel depth and air draft against the deepest and tallest vessel it must deliver, and a dock or launchway sized to that vessel. The sites left are compared on the rest: contiguous land with room for covered steel and outfit halls, crane capacity over the erection area, firm power, heavy-haul and water access, a trade workforce and training within reach, and the permits a waterfront needs.
Leaves for the rungs above, once its gate opens
One institution on one governed state · An approval path with class and regulatory authorities
Partners it needs, by kind
Classification societies and the flag-state authority · Suppliers of long-lead equipment · Workforce and training institutions · Yards and fabricators, where they are better · Port, waterfront and permitting authorities for a first site
Main risk
Capital outruns the evidence base. No land, workforce or supplier commitment is made until the units beneath have repeated with measured results, and no site is leased or bought before a program the evidence supports needs one. Partners carry capacity where they are better, so the company owns only what owning has been shown to improve.
Rung 6
Vessels
Long-termCapital phase: Production capabilityGate: proof level 8, vessel delivery
Difficult vessels designed, built, tested and delivered, each hull built better than the last because the evidence from the one before came back.
Stands on
- One institution on one governed state rung 5
- An approval path with class and regulatory authorities rung 5
- A decision record with named owners rung 0
Exists
Nothing yet.
Does not exist
A vessel
A vessel contract
A vessel design (the concept vessel is a synthetic schematic, not one)
People
Program management · Test and commissioning · Class and regulatory engagement · Fleet support
Capital
Vessel program working capital · Long-lead equipment · Test and trials · Class and regulatory approval
Equipment
Launch, test and commissioning facilities
Measured by
Build time and cost against the first hull · Defects at delivery · In-service availability
How it fails
Delivered late and over cost, like the programs it set out to improve · A configuration that cannot be approved for service
What would show it was rightA delivered vessel that works for a customer who depends on it, then a second built with less time and less rework than the first.
Rung 6: why here, the gate test, partners and risk
Why here
The vessel is the proof that the sequence was right, and it depends on every rung beneath it.
How the gate is judged
Judged by the customer in service, and by setting the second hull beside the first on the same basis: build time, cost and defects at delivery.
Leaves for the next hull
The first hull’s evidence, carried into the second
Partners it needs, by kind
A customer who depends on the vessel · Class and regulatory authorities for trials and approval · Fleet support providers
Main risk
Delivered late and over cost, like the programs it set out to improve. The first program is bounded and its design is mature before construction starts. Every decision on it is recorded against its evidence, so even a late or costly first hull tells the second one where the time went.

The destination

Build ships.

Saying it is not the same as claiming it. Each rung above must be earned first, in order.

What the sequence is built to compress

Cycle time is lost in specific places. Each mechanism below aims at one.

Intents, not results. No cycle-time figure is published until a physical unit has been measured against a baseline.

The 11 mechanisms and the loss each one targets

Design mature before production startsAimed at: late change discovered in steel
Tighter configuration discipline and earlier change-impact visibilityAimed at: work started against a stale revision
Material complete and accepted before work is releasedAimed at: shortages found at the work face
Modular construction with more outfitting done before erectionAimed at: outfit work in cramped, late, expensive positions
Supplier and long-lead visibility tied to the planAimed at: components that arrive after the work that needs them
Constraint-aware scheduling and digital work instructionsAimed at: crews assigned to work that cannot start
Rapid metrology and faster evidence returnAimed at: defects found blocks later, when fixing them costs most
Higher first-pass yield and tighter process controlAimed at: rework
Automation for suitable repetitive or hazardous workAimed at: throughput limited by scarce skilled hours
Clear release authority and shorter feedback loopsAimed at: decisions waiting on someone to find out who decides
A repeatable vessel platform, changed one section at a timeAimed at: every hull treated as a first of class

Outside evidence, not ours Leading commercial builders finish a design before construction and build on proven ones; the Navy falls short on these practices. GAO, May 2024. It validates nothing on this site.

Software first. Then steel. Then ships. The sequence to an American shipbuilder.

Governed build state

Operational reasoning

Program-wide governed state

Bounded physical cell

Integrated production unit

Integrated shipbuilding capability

Vessels

Software first. Then steel. Then ships.
The sequence to an American shipbuilder.