Your deployment, your rules.
EU by default. Sovereign by choice. The platform runs as SaaS in Frankfurt today; hybrid, on-premise and sovereign topologies arrive on Kubernetes at GA — from a Paris-headquartered vendor.
Today
Residency without negotiation
The loudest agent offerings anchor you to someone else's cloud. For European institutions with residency and sovereignty requirements, that is a constraint — so we made the opposite default.
EU (Frankfurt) by default
The platform region is not a premium tier or a negotiation. Deployment in the EU is the starting point, live today.
Your tenant boundary holds
Client business data stays inside your tenant boundary — enforced as row-level security on every table, not as a policy document.
A European vendor
NeoXam is headquartered in Paris, majority-owned by Eurazeo, with 35 years operating clients' most critical investment workflows.
Topologies
Four ways to run it — with honest dates
One Kubernetes-based architecture underneath all four. We say 'at GA' wherever it applies, because design intent is not a live capability.
SaaS — EU (Frankfurt)
ShippedThe default today. The platform runs in the EU; your business data stays inside your tenant boundary, isolated by row-level security.
Hybrid
GA Q3 2026Platform services in our EU SaaS, sensitive workloads and observability inside your perimeter. Kubernetes-based packaging.
On-premise
GA Q3 2026The full platform deployed on your Kubernetes, operated inside your boundary — for institutions whose policies end at their own racks.
Sovereign
GA Q3 2026Packaged for sovereign and regulated topologies: self-hosted observability, hardened secrets backends, no dependency on a US hyperscaler.
Data residency
Stated precisely, including the Beta nuance
Residency claims are worthless when they blur the edges. Here are ours, with their exact status.
EU (Frankfurt) deployment is live today. Business data stays inside the tenant boundary; secrets stay in vaults; no prompt or completion content reaches telemetry — contract-tested.
During Beta, trace storage may use a managed observability service under a disclosed, opt-in scope — never client business data — and moves fully inside the deployment boundary before GA.
Per-client region pinning, self-hosted observability inside your boundary, on-premise and sovereign packaging, hardened secrets backends.
GA Q3 2026
What general availability adds
The GA release is the multi-tenant, deploy-anywhere milestone. Early Adopter Program participants shape and validate it.
Per-client region pinning
Data residency committed per client, not per platform — pin your tenant to the region your regulator expects.
Per-tenant LLM keys & billing
Your own model API keys and clean billing attribution per tenant.
Scripted onboarding
Client onboarding scripted to complete in under a day.
Hardened secrets backends
HashiCorp Vault and AWS Secrets Manager support, with DR and backup procedures.
FAQ
Frequently asked questions
Where does the platform run today?
As SaaS in the EU — Frankfurt region — by default. Client business data stays inside the tenant boundary, enforced as row-level security in the database. NeoXam is headquartered in Paris and majority-owned by Eurazeo, a European investor.
Can we deploy on-premise or in a sovereign setup?
At GA in Q3 2026. The architecture is Kubernetes-based and designed for cloud, hybrid, on-premise and sovereign topologies, with a self-hostable observability stack. It is not available today, and we do not present it as if it were — Early Adopter Program participants prepare these deployments with us.
Does our business data ever leave our tenant boundary?
No — client business data stays inside your tenant boundary. One disclosed nuance: during Beta, trace storage may use a managed observability service under an explicit, opt-in scope that never includes client business data, and it moves fully inside the deployment boundary before GA.
Can we pin our data to a specific region?
Per-client region pinning is committed for GA in Q3 2026. Today the platform runs in the EU (Frankfurt) by default for all tenants.
Which model serves the agents, and where does that traffic go?
Agents run on Anthropic Claude through an AI gateway today. The model is a configuration field per agent, the harness is multi-provider by design, and per-tenant model keys arrive at GA — so model choice and routing follow your constraints, not ours. We never fine-tune models on your data.
Keep exploring
More of the platform
- Platform overviewThe governed agentic platform, end to end.
- Agents & catalogDeclared, versioned agents — descriptors, skills, modes.
- Governance & evalsPR review, eval baselines, allow/ask/deny policies.
- Security & trustIdentity, tenant isolation, vaults, append-only audit.
- Integrations & MCPMCP tools, NeoXam products, open standards, models.
- Observability & costOpenTelemetry traces, budgets, cost per run and tenant.
- ArchitectureRunner, harness, control plane vs execution plane.
General availability comes in Q3 2026. The Early Adopter Program is open now.
A limited cohort, a one-year platform trial and three workshop streams — Business ROI, Compliance, Operational fit. Bring one workflow; leave with a governed agent and the evidence to certify it.