See exactly where your data goes
Most AI vendors hand-wave on data. We show you the flow, the organisation isolation, what runs in our standard managed deployment, and where customer-specific regional controls apply — before you ever book a call.
Our promise: we do not use your organisation's content, conversations or learner data to train shared or third-party models. Improvement contributions are opt-in, per-organisation, and revocable at any time.
The same governance model is used across named institutional deployments and documented pilot outcomes.
4 — Institutional deployments · 400+ — AI characters across institutional cases · Web · LMS · Unity VR — One runtime across channels · Mar 2026 — Published JAMK pilot outcomes
How data flows
In our standard managed deployment, most core processing is pinned to EU regions where configured (for example Azure West Europe and Mistral Paris), with strict per-organisation scoping. Some optional provider paths or customer-requested integrations may involve processing outside the EEA; where that applies, we use DPA-based safeguards and applicable transfer mechanisms. The improvement path (dashed) is a separate, opt-in branch that is switched off by default for education and public-sector data.
Opt in or out — you stay in control
| Data | Default | Your control |
|---|---|---|
| Operational processing | Always on — it is the service | Required to answer & score interactions |
| Learner progress & analytics | On — your organisation owns it | Admin can export or purge at any time |
| Using your content to improve shared models | OFF by default | Explicit opt-in only · per-organisation · revocable |
| Voice audio retention | Transient by default | Configurable retention window |
The default posture above is a contractual commitment from day one. A self-serve toggle in your organisation's privacy settings is on our near-term roadmap; until then your data team controls it with us directly.
A ready product does not mean an integration-free product
SIELUNE is a standardised, licensed software product with a production AI runtime, knowledge-base system, administration panel, role-based assistant framework, API layer, user management and support model. As with all enterprise SaaS, each customer configures its own identity management, data sources, roles, permissions and integrations. This implementation work does not require redevelopment of the SIELUNE core product.
| Standard product (ships ready) | Status | Configured for you | Status |
|---|---|---|---|
| AI runtime (multi-model routing) | Production | Selection of your data sources | Configured during implementation |
| RAG knowledge-base system | Production | Document indexing | Configured during implementation |
| Knowledge Base Builder | Production | Roles & permissions | Configured during implementation |
| Administration panel | Production | SSO / HAKA / OIDC connection | Roadmap |
| Role-based assistants | Production | Moodle environment configuration | Configured during implementation |
| API layer (REST + WebSocket) | Production | M365 tenant authentication | Configured during implementation |
| Logging & analytics | Production | ServiceNow process mapping | Configured during implementation |
| Licence & support model | Standard module | Acceptance testing & capacity sizing | Configured during implementation |
Integration catalog — honest status
Moodle is open-source (GPLv3) and runs in thousands of institution-specific variations — different versions, plugins, themes and identity setups. By design there is no single integration that fits every Moodle, and even Moodle's own ecosystem is delivered through customization partners. Likewise, LTI is an open standard, not a product. That is exactly why enterprise SaaS always needs a configuration step. The difference with SIELUNE: the platform is one authenticated API, and our customization pipeline is already built and proven — we connected it live to a partner school's Moodle. Your integration is configuration on a ready pipeline, not a development project.
Because every capability is exposed through one authenticated API, M365, ServiceNow and federated SSO are delivered as customer-specific integration packages during onboarding. A packaged self-install native LTI plugin remains on the roadmap.
Standard enterprise SaaS implementation
- 1Tenant & security setup
- 2Identity & access configuration
- 3Knowledge-source connection
- 4LMS & system integrations
- 5Role & assistant configuration
- 6Acceptance testing
- 7Production launch
First production deployment can typically be completed in 6–8 weeks, depending on your identity environment, data-source readiness, integration scope and acceptance process.
Residency, GDPR & AI Act
Procurement-ready, with nothing hidden
Book a live demo or request the procurement pack — product description, architecture, DPA, security white paper, SLA and AI Act transparency statement in one place.