Products

Buy the process that hurts. Add the rest when it pays.

Hullira ships as five products mapped to the yard, not to abstract capabilities. Weldra controls and inspects welding, Fairix predicts and corrects distortion, Coatiq blasts and coats to spec, Aligna erects and aligns the blocks and hosts the twin, and Dockos optimises the dock against the launch date. Start with one on the Process tier and expand as each proves its number.

5 productsPer-process licensingWhole-yard bundle

run_4b7e19 · agent graph · block B-214 · scenario 10 nodes · 1 approval gate
01 · orchestrator ingest.package ✓ SUCCEEDED 02 · aligna twin.simulate ✓ SUCCEEDED 03 · cut_profile cut.nest ✓ SUCCEEDED 04 · panel_weld weld.sequence ✓ SUCCEEDED 05 · orchestrator approve.human ◆ APPROVAL 06 · quality_anomaly inspect.fuse ✓ SUCCEEDED 07 · fair_form fair.lineheat ✓ SUCCEEDED 08 · blast_coat coat.window ✓ SUCCEEDED 09 · schedule_yard dock.reoptimise ✓ SUCCEEDED 10 · erect_align erect.align ✓ SUCCEEDED

Illustrative scenario: Block B-214 · double-bottom block · panel line PL-2 · dock D1, one block, one approval gate

The five

One stack, five products, one record

Weldra controls welding and sees every weld defect by vision, X-ray and UT. Fairix predicts and corrects the weld-induced distortion those welds cause. Coatiq blasts and coats the hull to PSPC and class spec. Aligna erects and aligns the mega-blocks to millimetre tolerance and hosts the as-built ship twin. Dockos optimises dock, crane and berth schedules against the launch date. Every product feeds the others.

Weldra · panel and weld

Autonomous welding that hits class-grade quality without rework.

Weldra is the yard-edge weld-control and weld-inspection layer.

Fairix · fair and form

Predict weld distortion and fair the hull before it costs dock days.

Fairix predicts and controls weld-induced distortion.

Coatiq · blast and coat

Blast and coat to PSPC and class spec the first time, every square metre.

Coatiq is the blasting-and-coating autonomy layer.

Aligna · erect and align

Erect and align mega-blocks to millimetre accuracy, simulated before the lift.

Aligna controls block erection and dimensional accuracy.

Dockos · schedule and yard

Optimise dock, crane and berth schedules against a hard launch date.

Dockos is the yard scheduling and logistics optimisation layer.

Weldra · ■ brand #0C6EB4

Weldra

Autonomous welding that hits class-grade quality without rework.

The problem

A single vessel carries hundreds of kilometres of welds, and one drifting parameter, heat input, travel speed, wire feed or fit-up gap, silently produces porosity, lack of fusion or cracks that an X-ray or UT surveyor only catches days later, after the block is closed. By then a class-rejected weld means grinding out, re-welding, re-inspecting and a slipped dock date that rents for six figures a day. Welders skilled enough to hold the arc steady are retiring faster than yards can replace them, and periodic inspection leaves managers blind on real-time weld quality across the panel line and block stages.

What Weldra does

Weldra is the yard-edge weld-control and weld-inspection layer. Weld-pool, seam and bead cameras plus X-ray and UT streams run fine-tuned models that classify defects and predict defect risk in real time, while a control agent adjusts weld parameters and sequences for robotic and manual-assist welding to hold class-grade quality without distortion. Every weld is logged with WPS and PQR evidence to an assurance-grade audit trail. Weldra emits a live weld-quality signal that Fairix and Aligna consume.

Who it is for

Commercial newbuild yards building tankers, bulkers, containerships and gas carriers, naval and defense shipbuilders, and repair and conversion yards; welding-robot and CAD/CAM OEMs embedding Hullira-ready weld autonomy; multi-yard shipbuilding groups.

Key features

  • Real-time weld-pool, seam and bead defect vision plus X-ray and UT classification
  • Adaptive weld-parameter and sequence control for robotic and manual-assist welding, inside the WPS envelope
  • Defect-risk prediction ahead of the surveyor, in time for an approved corrective move
  • WPS and PQR cited, immutable weld-quality audit trail per joint
  • Feeds Fairix and Aligna with a common weld-quality signal

NVIDIA stack

Named in the product brief as the design

Jetson OrinDGX / HGXHoloscanDeepStream / MetropolisTensorRTIsaacTritonNVIDIA AI EnterpriseNIM / NeMo

Planned

Jetson Thor for high-frame-rate multi-camera cells · plannedFederated rare-defect transfer across yards · planned

Why it needs a GPU

Real-time multi-camera weld inference at 15 to 60 FPS with sub-50 ms seam-alert latency cannot run on CPU, and training weld-defect and X-ray or UT models on millions of frames plus synthetic rare-defect variants needs DGX or HGX class hardware.

MVP scope and first metric

Retrofit one welding cell with weld-pool cameras and a UT feed; ship defect classification and risk alerts as advisory, then graduate to parameter and sequence control hand-off with class-gated human approval. First metric: weld rework or class-rejected welds on a named panel line.

Tier. Process tier, $12,000 per welding process or area per month, expanding to Yard and Enterprise.

Open questions

  • Per-cell camera counts, frame rates and UT sampling vary by yard and process
  • Federated cross-yard defect transfer is planned and waits on data and IP agreements
Fairix · ■ brand #0C6EB4

Fairix

Predict weld distortion and fair the hull before it costs dock days.

The problem

Welding pumps heat into steel, and steel moves. A panel that measured true before welding warps by millimetres after, and those millimetres compound across sub-blocks and blocks until a hull section will not fit its neighbour at erection. Yards fight this with manual line heating and mechanical fairing, slow, craft-bound work done by a shrinking pool of shipwrights, plus periodic laser scans that reveal the distortion only after it is baked in. Every out-of-tolerance block means fairing labour, re-measurement and days lost in a dock renting for six figures a day.

What Fairix does

Fairix predicts and controls weld-induced distortion. A GPU-accelerated weld-thermal and distortion FEA surrogate forecasts how each block will move given weld sequence, heat input, material and geometry, then a fairing agent recommends and controls weld order, back-stepping and line-heating moves to hold dimensional tolerance. Laser-scan and photogrammetry point clouds confirm as-built against as-designed in near real time. Fairix takes Weldra's live weld-quality signal as input and hands Aligna a distortion-corrected block, validated in the yard twin before the shift.

Who it is for

Commercial newbuild, naval and repair or conversion yards with accuracy-control teams; hull and accuracy-control engineers; welding-robot and laser-scan OEMs embedding distortion prediction.

Key features

  • GPU-accelerated weld-thermal and distortion FEA surrogate for pre-shift distortion forecasting
  • Weld-sequence, back-step and line-heating recommendations to hold tolerance
  • Laser-scan and photogrammetry point-cloud comparison of as-built against as-designed
  • Fairing-move logging with dimensional-accuracy evidence for class
  • Consumes Weldra weld data; feeds Aligna a distortion-corrected block

NVIDIA stack

Named in the product brief as the design

H100 / HGXJetson OrinDGXModulus / CUDAOmniverseTensorRT / TritonNVIDIA AI EnterpriseNIM

Planned

Cosmos / Omniverse Replicator for rare distortion modes · planned

Why it needs a GPU

Weld-thermal and distortion FEA and structural simulation are compute-heavy. Running them fast enough to plan a shift, plus processing dense point clouds, is not possible on CPU; surrogate training and Omniverse block simulation need H100 or DGX class hardware.

MVP scope and first metric

Ingest the weld sequence and one block's scans, predict distortion offline, validate against as-built scans, then move to live fairing recommendations with accuracy-control-engineer approval. First metric: fairing hours or out-of-tolerance blocks on a named repeat block geometry.

Tier. Process tier, $12,000 per process or area per month for distortion and fairing control, expanding to Yard and Enterprise custom distortion models.

Open questions

  • Material grades and scan cadence vary by yard and block
  • Synthetic distortion-mode generation is planned and waits on twin fidelity work
Coatiq · ■ brand #0C6EB4

Coatiq

Blast and coat to PSPC and class spec the first time, every square metre.

The problem

A ship's coatings are what let it survive decades at sea, and yards blast and coat millions of square metres per vessel against IMO PSPC and classification-society spec. Get the surface prep, humidity window, dew point or film thickness wrong and the coating fails early, driving warranty claims, class findings and rework in ballast tanks and hull areas that are miserable and expensive to re-access. Today crews chase environmental windows by hand, spot-check dry-film thickness and inspect coatings periodically, so faults surface late and traceability for PSPC audits is patchy.

What Coatiq does

Coatiq is the blasting-and-coating autonomy layer. It fuses environmental sensors for humidity, dew point and surface temperature with surface-profile vision and dry-film-thickness readings to control when and how blasting and coating run, holding surface prep and coating thickness inside PSPC and class spec. A coating agent schedules work to the right environmental window and drives robotic blast and coat cells, while the conformance agent assembles PSPC evidence automatically. Coatiq protects the hull that Weldra welded and Aligna erected, and feeds Dockos the coating-window constraints that shape the dock schedule.

Who it is for

Commercial newbuild, naval and repair or conversion yards; coating applicators and QA or coating inspectors; coating-equipment and robotic blast and spray OEMs embedding Hullira-ready coating control.

Key features

  • Environmental-window control on humidity, dew point and surface temperature for blast and coat timing
  • Surface-profile vision and dry-film-thickness assurance to PSPC and class spec
  • Robotic blast and coat cell control with human-approved autonomy
  • Automated PSPC and class coating evidence and traceability
  • Feeds Dockos coating-window constraints; protects Weldra and Aligna hull work

NVIDIA stack

Named in the product brief as the design

Jetson OrinDGX / HGXDeepStream / MetropolisHoloscanTensorRTIsaacTritonNVIDIA AI EnterpriseNIM / NeMo

Planned

Jetson Thor for higher-throughput coating vision · plannedSynthetic coating-failure generation · planned

Why it needs a GPU

Real-time surface-profile and coating-anomaly vision across large hull areas, fused with environmental telemetry at the edge, cannot run on CPU; training coating models on large image corpora plus synthetic failures needs DGX or HGX class hardware.

MVP scope and first metric

Instrument one coating area with environmental and thickness sensors and cameras; ship environmental-window guidance and thickness alerts as advisory, then graduate to robotic blast and coat control with QA approval and automatic PSPC evidence. First metric: coating non-conformance or re-coat area on a named ballast-tank or hull area.

Tier. Process tier, $12,000 per process or area per month for coating control, expanding to Yard and Enterprise coating-fleet contracts.

Open questions

  • Coating systems, sensor suites and PSPC evidence formats vary by yard
  • Synthetic coating-failure generation is planned and waits on data collection
Aligna · ■ brand #0C6EB4

Aligna

Erect and align mega-blocks to millimetre accuracy, simulated before the lift.

The problem

Modern ships are built as house-sized blocks and joined in dock, where two mega-blocks weighing hundreds of tons each must meet to millimetre accuracy. A misalignment discovered mid-erection means re-cutting, re-fairing, re-welding, and crane and dock time burned at six figures a day, and every block that arrives already distorted makes the fit worse. Yards plan erection on drawings and check alignment with periodic laser scans and photogrammetry, so problems appear during the lift instead of before it, when they are cheap to fix.

What Aligna does

Aligna controls block erection and dimensional accuracy. It builds an as-built ship twin in Omniverse from laser-scan and photogrammetry point clouds, simulates block reach, crane paths and erection order, and predicts fit before the lift, then guides the alignment crew through positioning and shimming to hold tolerance. It takes distortion-corrected blocks from Fairix and weld quality from Weldra, and its erection schedule feeds Dockos. The twin becomes the yard's single source of as-built truth for class approval.

Who it is for

Commercial newbuild and naval yards running block and mega-block erection; accuracy-control and hull engineers; laser-scan, photogrammetry and crane-automation OEMs embedding erection accuracy.

Key features

  • As-built ship twin in Omniverse from point clouds, as-built against as-designed
  • Erection-order, crane-path and reach simulation before the lift
  • Block-alignment and shimming guidance to hold millimetre tolerance
  • Dimensional-accuracy evidence for classification-society approval
  • Consumes Fairix and Weldra output; feeds Dockos the erection schedule

NVIDIA stack

Named in the product brief as the design

DGX / OVXH100 / HGXJetson OrinOmniverseCUDAcuOptTensorRT / TritonNVIDIA AI EnterpriseOmniverse Cloud

Planned

Cosmos / Omniverse Replicator for misalignment cases · planned

Why it needs a GPU

Dense point-cloud registration, structural-fit computation and real-time Omniverse twin simulation of house-sized blocks and crane paths cannot run on CPU at useful speed. Twin rendering and photogrammetry need DGX, OVX or H100 class hardware.

MVP scope and first metric

Build an as-built twin from one block's scans, simulate a single erection join and predict fit, then move to live alignment guidance with accuracy-control approval. First metric: dock hours per join and re-cut or re-fair events on a named block join.

Tier. Process tier, $12,000 per process or area per month for erection accuracy, expanding to Yard and Enterprise multi-yard and naval-program twin contracts.

Open questions

  • Scan rig fidelity and crane telemetry access vary by yard
  • Synthetic misalignment generation is planned and waits on twin calibration
Dockos · ■ brand #0C6EB4

Dockos

Optimise dock, crane and berth schedules against a hard launch date.

The problem

A shipyard is paced by a handful of scarce, six-figure-a-day assets: the dry docks, the heavy cranes, the berths. A slip anywhere, a late block, a stalled weld, a coating window missed, a crane double-booked, cascades into the whole build programme and pushes the launch date. Yards plan this in spreadsheets and MES dashboards that monitor status but do not optimise, so managers react to slips instead of preventing them, and a single retiring planner often holds the yard's sequencing craft in their head.

What Dockos does

Dockos is the yard scheduling and logistics optimisation layer. It runs GPU-accelerated optimisation over dock, crane, berth, work-package and yard-logistics constraints, ingesting live progress from Weldra, Fairix, Coatiq and Aligna to re-optimise the schedule continuously and flag class and schedule risk before it bites. It turns the yard from a set of siloed status boards into one optimised, self-correcting plan that protects the launch date.

Who it is for

Commercial newbuild and naval yards and multi-yard groups; production directors and yard planners; yard-MES and scheduling vendors embedding Hullira-ready optimisation.

Key features

  • GPU optimisation of dock, crane and berth schedules and work-package sequencing
  • Live re-optimisation from Weldra, Fairix, Coatiq and Aligna progress signals
  • Yard-logistics routing for blocks, materials and transporters, with clash avoidance
  • Class and schedule-risk flagging with explainable, cited reasoning
  • Consumes the whole Hullira loop; produces the yard's single optimised plan

NVIDIA stack

Named in the product brief as the design

H100 / HGXDGXJetson OrincuOptRAPIDSNeMoTriton / TensorRTOmniverseNVIDIA AI EnterpriseNIM

Planned

What-if scenario simulation in the shared twin · planned

Why it needs a GPU

Optimising dock, crane and berth assets and thousands of work packages against constraints, re-solved as live progress changes, is a large combinatorial workload that cuOpt runs on GPUs. CPU solvers cannot re-optimise fast enough to act within a shift.

MVP scope and first metric

Connect one yard's dock, crane, berth and work-package data; run cuOpt scheduling in advisory mode, then close the loop with live re-optimisation from Hullira process signals under planner approval. First metric: dock utilisation and on-time block erection on a named dock.

Tier. Sold inside the Yard tier ($85,000 per month) and Enterprise multi-yard and group scheduling contracts as the scheduling layer.

Open questions

  • MES and scheduling data models and asset constraints vary by yard
  • What-if scenario simulation depth is planned and waits on twin integration
Compare

What is in each tier

Products are licensed per process on the Process tier and bundled on Yard and Enterprise.

What is in each tier
CapabilityProcessYardEnterprise
Processes covered1All in the yardAll in the group
Agent modules1 of 5 productsAll agentsAll agents + custom models
As-built ship twin (Aligna)NoIncludedCustom vessel-class twins
Dock optimisation (Dockos)NoIncludedGroup and program scheduling
Yard-edge nodeIncludedPer process areaFleet managed
Guarded controller writesScoped tagsAll scoped tagsGroup policy
Block genealogyIncludedIncludedIncluded + class evidence packets
Autonomy levelsL1 to L2L1 to L3L1 to L4 when accepted
SSO and RBACNoPlannedPlanned, with approval chains
On-prem deploymentNoOptionalIncluded
SupportBusiness hours24/724/7 + named engineer
Indicative price$12,000/mo$85,000/mo$700k to $10M ACV
Run timeline · scenario

The five products in one block

Every Hullira run is an ordered, inspectable sequence. This is run_4b7e19 on Block B-214 · double-bottom block · panel line PL-2 · dock D1. Scenario completes in 10 steps with one approval gate. Predicted distortion at the erection joint fell from 4.2 mm to 1.8 mm after a back-step sequence, one porosity precursor on seam S-31 was cleared inside the WPS envelope, the ballast-tank coat ran inside a 3.4 °C dew-point margin, and the block landed at 1.6 mm with its conformance record written.

run_4b7e19 Block B-214 · double-bottom block · panel line PL-2 · dock D1 illustrative scenario SUCCEEDED
  1. 01 ingest.package SUCCEEDED 0.8 s

    Yard Orchestrator · Pulled the work package for B-214 from the yard MES: 46 plates, 212 seams under WPS-117, erection tolerance 2 mm at the D1 joint, PSPC-grade ballast-tank coating, erection slot on D1 in the dock plan. Locked the qualified WPS envelope and the class scope.

  2. 02 twin.simulate SUCCEEDED 1 m 12 s

    Ship Twin · Simulated 18 candidate weld sequences on the as-built twin with the weld-thermal surrogate, ranked on predicted distortion at the erection joint, crane reach on D1 and shift takt. Standing sequence predicts 4.2 mm; candidate #6 predicts 1.8 mm.

  3. 03 cut.nest SUCCEEDED 3 m 40 s

    Cut-and-Profile · Released the nest for 46 plates on CL-1 with the fit-up gaps candidate #6 assumes. As-cut geometry logged per plate and handed to the panel line; material yield 91.4% against the nest plan.

  4. 04 weld.sequence SUCCEEDED 2 m 05 s

    Panel-and-Weld · Loaded the back-step sequence for the non-critical seams on PL-2 and set heat input and travel speed inside WPS-117. Weld-pool and seam cameras streaming on 12 channels.

  5. 05 approve.human APPROVAL 58 s

    Yard Orchestrator · Candidate #6 also re-orders two class-critical butt seams, S-31 and S-32. Sequence changes on class-critical seams stay with a human, so the step was held for the shift welding engineer and approved at 06:12:44.

  6. 06 inspect.fuse SUCCEEDED continuous

    Quality-and-Anomaly · Twelve weld cameras and the UT station fused at the yard edge. One porosity precursor on S-31 at 06:41 flagged and cleared by a 4% wire-feed trim inside the WPS-117 envelope; the joint was routed to UT and passed.

  7. 07 fair.lineheat SUCCEEDED 4 m 30 s

    Fair-and-Form · Post-weld scan registered against the twin: 2.1 mm at the erection joint. Two line-heating passes planned on the inner bottom and executed by the fairing crew; re-scan read 1.8 mm, inside the 2 mm tolerance.

  8. 08 coat.window SUCCEEDED continuous

    Blast-and-Coat · Ballast-tank blast and first coat scheduled to a window with steel temperature 3.4 °C above dew point. Dry-film thickness read 312 µm against a 280 µm minimum on the stripe coats; PSPC evidence assembled per area.

  9. 09 dock.reoptimise SUCCEEDED 41 s

    Schedule-and-Yard · Re-solved the D1 dock and crane plan with B-214 ready 6 hours early and a neighbouring block late: erection slot pulled forward one shift, crane G2 reassigned, no berth conflict. Planner approved the revised plan.

  10. 10 erect.align SUCCEEDED 1 h 48 m

    Erect-and-Align · Both blocks scanned on D1 and registered against the twin; fit predicted at 1.6 mm. Shimming guidance issued to the erection crew, block landed and welded out, final scan 1.6 mm. Conformance record written: welds, scans, coating readings, approvals and class evidence for B-214.

Accelerated computing

Physical AI, at the arc

A weld pool is decided in milliseconds and a dock waits by the hour. Perception has to be local, deterministic and fast, so Hullira runs GPU inference at the yard edge and keeps training, simulation and optimisation in the cloud or on-prem.

Jetson Orin · DeepStream · Holoscan · TensorRT

Edge perception

6 to 24 cameras per process area plus UT, X-ray, thermal, laser-scan, photogrammetry, robot and coating-instrument streams fused at the yard edge. Sub-50 ms weld alerts and sub-150 ms coating or scan prompts are the design targets. Jetson Thor is planned for higher camera density.

Triton · NIM

Model serving

Weld-defect, distortion-risk, coating-risk, point-cloud and schedule-risk models behind one orchestrator at the yard edge, versioned with rollback, routed by latency, tenant, sensitivity and cost.

DGX or HGX · NeMo

Training

Weld-defect vision, UT and X-ray classification, NeMo class-rule reasoning and weld-thermal FEA surrogates fine-tuned per tenant: 100 to 500 GPU-hours per yard during onboarding, then monthly retraining from engineer corrections and class outcomes. Dedicated training capacity is planned.

Omniverse · OVX · Modulus

Ship and yard twin

Block erection, weld sequence, robot reach, crane paths and yard layout simulated before the shift, as-built against as-designed. GPU weld-thermal and structural FEA is the workload; physics-informed surrogates rank plans at shift cadence.

Cosmos · Omniverse Replicator

Synthetic data

Rare weld defects, coating failures, distortion modes, occluded seams and erection misalignment cases generated and validated against real NDT and scan distributions before promotion. 50k to 250k variants per vessel class is the target. Planned.

cuOpt

Optimisation

Dock, crane and berth schedules, work-package sequencing, yard logistics and erection order, re-solved as progress changes.

Isaac · Isaac ROS

Robot cells and crawlers

Robotic weld, blast, coat and inspection cells and crawler or drone hull inspection, with paths validated in simulation before they run next to a hull.

RAPIDS

GPU ETL

Telemetry ETL across weld parameters, scans, NDT results, coating readings, non-conformance and schedule history, for retraining, drift detection and rework, dock-time and schedule benchmarks.

NeMo Guardrails · Retriever

Policy and grounding

WPS and PQR, class-rule, PSPC, cyber and human-approval policies enforced before any action writes back to robots, MES, QA or work packages; retrieval with citations and tenant isolation.

Pricing

Priced per process, per yard, per group

Land on one workflow with a measured baseline. Expand by process, yard, group site and autonomy level. Outcome components are available on rework, dock time and schedule value. Annual prepay carries a 15 to 20% discount.

Billing period
Process $12,000 per production process or area / month · billed monthly

Weld, fairing or coating control, or weld-inspection AI, for one process. For yards validating ROI on the wedge.

  • One process or area, one agent module
  • Yard-edge inference node included
  • Robot, instrument and MES reads, guarded writes
  • Run log, audit trail, block genealogy
  • Business-hours support
Start a pilot
Enterprise Custom land $700k to $10M ACV

Multi-yard groups and naval programs standardising on Hullira, with custom weld, distortion and schedule models.

  • Multi-yard and program fleet management
  • Custom weld, distortion and schedule models
  • Yard-edge fleet with on-prem option
  • Group autonomy policy and approval chains
  • SLAs, SSO and RBAC, dedicated engineer
Contact us
FAQ

Straight answers

The questions production directors, welding engineers and class-compliance leads ask in the first meeting.

Yes, but only within an explicit tag allow-list with per-tag rate and magnitude limits, inside the qualified WPS, coating, dimensional and schedule constraints, and only at the autonomy level your yard has set. Design partners start in shadow mode, where Hullira writes nothing. Class-critical seam changes, lift plans and block release stay with named people.

Get started

Pick your first product

Start with one welding, fairing, coating or erection process and one measurable baseline. A paid pilot on weld inspection, distortion control or coating assurance shows the number before you commit further.