- Material
- Labor
- Freight cost
- Footprint
Measured, not assumed.
Plan early. Predict with confidence. Reconcile with provenance.
CubeRight is a predictive cubing engine: cartonization that resolves the whole physical order the day it lands — cartons, cube, weight, pallets and freight class, days before any of them are committed. Any shape, either freight mode, and the measurement behind every number stored beside it.
Every figure on this page was produced by executing the engine
What it costs to guess
The commitments come due before the cube exists.
Board runs in weeks. Carriers in weeks. Labor and dock slots in days. Every one is priced in cartons, pallets or density — and every one is committed before the order is cubed. CubeRight resolves the cube the day the order lands. Everything downstream is then priced on a measurement instead of an estimate.
↔ Scroll the timeline
Cubing at the pack bench is the immature case, and we do not claim it is typical. Most operations resolve cube at wave; the mature ones at tender, because the tender itself asks for cartons and pallets. Both are real capability.
Freight class is declared before the cube exists.
Current operating options
Four capable options, all of them late.
Each is a reasonable purchase, and several pack extremely well. The objection is not quality. It is that all four answer the question after the expensive decisions have already been made.
Makes a better box, cut to the goods. Genuinely reduces void — and it makes that box at the pack bench, which is exactly where the information was already too late.
Answers atthe bench · minutes before ship
Already paid for, and it runs in the warehouse at wave planning. It is pseudo-cubing: it optimizes for the fewest boxes, which is not the same objective as the cheapest shipment — and wave is still hours, not days.
Answers atwave · 2–8 hours out
Packs well, returns fast, and integrates in an afternoon. It hands you a box choice rather than a plan — and it is called by something that already decided when to ask.
Answers atwhenever it is called
Broad, capable, and a six-to-twelve-month programme — the TMS and ERP end of the stack. It starts from shipments rather than from orders, which puts its first useful number downstream of the commitments on the list above.
Answers atshipment · after the programme
The product
A running system that shows its work.
Not an endpoint you integrate and hope about. Every solve keeps its own audit trail, the 3D viewer is how you read it, and savings are reported as the measured difference between two real runs. The number on your dashboard is the one we would defend in a room.
Persistence
Multi-tenant, migrations, audit log, rotating backups, integrity checks.
Interface
REST surface with a drift-checked catalog, CSV import/export, API tokens.
Access
RBAC, tenancy isolation, full audit trail on every solve.
Application
Browser SPA with a 3D viewer and in-app documentation.
Shape coverage
Cuboid, cylinder, sphere and conformable — native, not bounding-box.
Labeling
SSCC / GS1 hierarchy and LPN, generated with the pack.
Analytics
Measured policy baselines in raw physical units. Prices applied at read time, never stored.
Constraints
Orientation, bearing and crush, weight caps, fill and void, carton eligibility, receiver height caps.
Where the money actually is
The money is in trailers, not cardboard.
The industry advertises corrugate because it is the biggest-looking percentage available. Measured, freight leads it by roughly nine to one. And trailer utilization is lumpy rather than linear: pallets and trailer positions are integers, so a gain in cube utilization is worth nothing until it crosses a threshold and then it is worth a whole trailer. That is the honest reason small pilots underwhelm, and we would rather say so before one than after.
+50% on the LTL rate per pallet swings total value
+36%
+50% on corrugate $/m² swings total value
+4.0%
+50% on burdened labor rate swings total value
+2.3%
Spread between an LTL lane and a TL lane, same physical improvement
2.0×
Measured on 200 synthetic grocery orders, re-run 28 August 2026: optimized policy against the product's own naive baseline, palletised identically under both, identical units and identical weight in each run. The shares above are the more conservative of two objectives. Asked for the fewest pallets the engine returns freight at 71.4% and corrugate at 8.1%; asked for the fewest cartons, 75.2% and 6.5% — so freight leads corrugate by between nine and twelve times, and the figure quoted is the lower one. Value lands between $17 and $20 per order on an LTL lane and $8–$9 on truckload, or $1.7M–$2.0M a year at 100,000 orders. Mode determines everything: qualify the LTL/TL mix before anyone quotes a number. The full delta table, the controls and the caveats are on the evidence page.
Basis. 200 synthetic grocery orders, one lane, measured 28 August 2026 — not a customer book. Freight class is indicative: this run produced no NMFC class shift, though earlier runs did, and a class change is worth real money when it happens without being something you can count on.
Proprietary · patent pending
The shape itself is part of the answer.
Every cubing engine on the market treats an item's dimensions as fixed and packs around them. Where the shape is genuinely negotiable — wire, cable, hose, rope, tubing — CubeRight decides it.
Same material length in every form. Only the geometry changes.
A fixed length of cable can be wound wide and flat, or narrow and tall. The volume never changes; the right form depends on what it is going into.
We tested thirteen products across every category that touches this problem. Every one takes item dimensions as immutable input.
B Provisional filed with the USPTO, August 2026. Patent searching returns mechanical art only — winding machines, spools, drums. No software prior art for choosing packed geometry to optimize a shipment. The full argument, and the three claims we refuse to make.
The one ask
We measure the saving. We don't assume it.
Most cubing ROI models multiply an assumed percentage by your freight spend. We would rather run your own book both ways and hand you the difference in physical units, priced with your rate card.
Twenty minutes, not a pitch
The engine on real data, and your edge cases answered. No deck.
Then we run your book, both ways
Your policy against the optimized one. No integration, no systems work.
And we tell you if it is small
If you already cube well the difference is modest. There is no price on this step.