Land low-friction on one workflow with clear ROI, expand across cells, lines, modules, sites and autonomy level, and monetise the value with usage and outcome components as the agent does more.
Built with foundries and die casters melting iron, steel, aluminium, magnesium and copper
[PLACEHOLDER] Design-partner names are illustrative until first references are signed.
Every tier runs the same factory-edge runtime and the same autonomy gates. What changes is scope, model customisation and commercial structure.
Foundries validating ROI on the wedge.
$15,000 /cell/mo
billed monthly
Most popular
Scaling plants adopting the full loop.
$90,000 /mo
billed monthly
Foundry groups standardising on Smelteon.
Custom $650k–$8.5M ACV
billed monthly
Annual pricing shown with the 18% prepay discount applied. Usage above plan limits is metered per casting, per cell or per action.
Shadow, assist and bounded modes with human-in-the-loop checkpoints on every tier. Nobody buys their way past a safety review.
Assurance-grade logging of every agent action, approval, override and rollback — included, not an add-on.
Your recipes, geometries and telemetry are isolated on every tier, with on-prem available where required.
The console metallurgists and process engineers use to approve, correct and train the models — the core of the loop.
| Capability | Cell | Factory | Enterprise |
|---|---|---|---|
| Scope | One cell or line | Whole plant | Multi-site group |
| Agents | One | All nine | All nine plus custom |
| Castwin twin | Plan consumption | Full simulation | Group-wide, custom families |
| Model tuning | Shared priors | Plant-tuned | Custom alloy / geometry / tolerance |
| Edge footprint | Cell node | Plant fleet | Managed multi-site fleet |
| Genealogy & conformance | Audit log | Full genealogy | Group assurance packages |
| Support | Standard | Priority | SLAs, named contacts |
| Outcome pricing | — | Optional | Standard on top workflows |
| Term | Monthly or annual | Annual | Multi-year with ramps |
We target wedge cells with $500k–$5M a year in scrap, rework, melt-energy loss, inspection bottleneck or labour constraint. Shadow mode measures it against your own records rather than an estimate.
78%
Target gross margin at scale
135%
Target net revenue retention
$3.6M
Target LTV
$90k
Blended enterprise CAC
Business-model targets. Primary COGS is inference and compute, offset by routing high-volume steps to fine-tuned open models, caching and distillation.
One avoided escape pays for the year. Everything after that is yield.
We priced it against the two pourers we cannot hire, not against a software line item.
Outcome pricing on porosity was what got finance to stop arguing about the subscription.
Design-partner voices are illustrative composites of foundry buyer conversations. [PLACEHOLDER — replace with named references after first case studies.]
One melt, mold or pour cell or line — the unit a foundry already manages as a thing. Inspection lines count as a cell. If you are unsure how your plant maps, we will scope it during the plant walk.
Because the highest-value workflows — yield, scrap, porosity and first-pass yield — produce measurable improvements against a baseline we agree in advance. Sharing that upside aligns our incentives with the number you actually care about.
Floor pricing is protected. Discounts are traded for term length and case-study rights, plus a 15–20% annual prepay discount to improve cash and retention on both sides.
A pilot is a paid engagement on the Cell tier against a defined success metric. We would rather charge for a pilot that proves a number than run a free trial that proves nothing.
The target is under 6–12 months on a cell where scrap, rework, melt-energy loss, inspection bottlenecks or labour constraints cost $500k–$5M a year, and under 14 months blended at enterprise scale.
Expansion is the primary growth engine and it is priced to be easy: add cells, modules, lines or sites, with usage metering above plan limits and enterprise terms once the group standardises.
We will scope the wedge, agree the success metric, and price the pilot against it.