The European Foundry for regulated industry
A foundry is a place where raw material is forged into finished product. In our world, that means a platform where your operational data, from MES, ERP, lab and historian, fuses with AI agents, typed actions and audit trails to produce operational action, not just analytics.
What we mean by “Foundry”
Foundry is our category metaphor, not our product name. Palantir Foundry has proven the category globally. We are the European alternative, with sovereignty, honest scoping and phased pilot as our differentiators.
Your existing data sources
Process Historian (PI/AVEVA), LIMS, ERP (SAP/Monitor/IFS/M3), MES, SCADA. We don't replace them, we observe them and write back with audit.
The Melt
Ontology runtime + AI agents + audit pipeline + role-based ACL, all on a shared truth. One semantic layer between your existing systems and the people, agents and dashboards that consume them.
Not just analytics
AI-proposed work orders. Audit trails end-to-end. Real-time deviations. Visual reasoning rendered inline. CSRD and EU AI Act evidence built in.
Foundry is not our product name. Our product is AEL Studio, the Agentic AI platform with ontology runtime, AI mesh, and built-in compliance. Foundry is the category we position AEL Studio in. When customers say “we’re evaluating Foundry-class platforms”, they mean Palantir Foundry, AEL, and a handful of others. We’re the European, sovereignty-first alternative in that category.
AEL Studio + Ontology Runtime + your stack
The semantic layer that sits between your existing systems and AEL’s AI mesh. All audit-logged, runs in your own environment, on-prem or in the EU region. We don’t replace your source systems, we observe them and write back with audit and approval gates.
Three lanes: users and interfaces on top, AEL Studio with the Ontology Runtime in the middle, customer systems at the bottom. Green arrows are audited writebacks, dashed arrows are deferred capabilities brought in on customer demand.
Open with legend in new tab →Four pillars that define our approach
Each pillar is a deliberate design choice, and a deliberate departure from how the global incumbent operates. They’re what make the European Foundry category different from “AI dashboards” or a low-cost Palantir Foundry clone.
European sovereignty by design
Swedish legal entity. EU jurisdiction. EU region or on-prem. No Schrems II, no CLOUD Act exposure. Your operational data, quality deviations, production conditions, customer relationships, is competitive shielding. We won't put it where you don't control it.
Sovereignty as architecture
OSS stack on .NET Aspire + Kubernetes. Customer-owned Git repository extending our component repos. Peer-reviewed CI/CD. Customer-pulled Docker / K8s updates via Docker Hub. AEL retains copyright on the runtime; you receive a permanent MIT license. If we disappear, you keep going.
Honest scoping, no FDE black box
Phased pilot with GO/NO-GO between phases. Fixed price per phase, no sprawling consulting. Two to four people for 8 to 12 weeks, not 80 forward-deployed engineers for 12 months. Partner-led rollouts beyond pilot.
Model-agnostic AI mesh with visual reasoning
Anthropic, OpenAI, Mistral, local models, your fine-tunes, switchable without rewriting systems. Audit and observability follow regardless of model. Mermaid graphs and 5 Why analyses rendered inline by the agent, visual reasoning your operators understand in 5 seconds, your auditors accept as evidence.
Compounding ROI, every use case after the first costs less
Without an ontology, every new use case is a new data model, a new integration, a new pipeline. With a Foundry-style ontology runtime underneath, use case 5, 10, 20 reuse the same objects, actions and audit infrastructure. Your marginal cost per use case drops 50 to 70 % after the first.
“We’re not selling a feature list. We’re selling compounding. The first use case costs full price. The fifth costs a fraction. By the twentieth, your marginal investment is configuration, not engineering.”
Honest about where we fit
Not every AI use case benefits from a Foundry approach. We’re explicit about that, because honest scoping is one of our pillars.
| You should consider Foundry when… | You probably don’t need Foundry when… |
|---|---|
| You have multiple data sources (ERP, MES, historian, lab) that don't share a definition of “the same thing” | You have one cleanly modelled data source and one well-defined task |
| You expect to add use cases over time and don't want to rebuild the data model each time | You're solving a single, isolated automation problem |
| You need audit trails for CSRD, EU AI Act, ISO 9001/14001 or sector-specific compliance | Compliance is not a primary driver |
| You're considering Palantir Foundry but the price, lock-in or US jurisdiction is a problem | You're happy with a SaaS in someone else's cloud and don't need EU sovereignty |
| You want AI agents to write back to ERP / SAP PM with proper governance, not just read | You only need read-only analytics |
Want to see what a Foundry pilot looks like in your industry?
Our 12-minute live demo walks through three acts: a deviation discovered earlier than today, an AI-proposed work order with full audit trail, and a brand-new compliance parameter added live in 90 seconds. We can run it on a generic example or, with two weeks of preparation, on a sketch of your actual domain.