One seamless application
Four delivery modes. One record spine. One transcript.
“One integrated application” is a contractual definition here, not a marketing adjective — and it is enforced by a test that runs on every build.
The acceptance test we hold ourselves to
A learner completes a knowledge module on the web, a virtual classroom session, a practical simulation, and an offline workplace logbook entry on a phone — with zero visible context switches and a single unified transcript at the end. That journey is an executable acceptance test in our suite. It passes. When it stops passing, the build stops.
- One identity, everywhere — including content launches, virtual classroom joins and headset sessions
- One design system across web, mobile and immersive menus
- One navigation shell
- One learning-record spine
- One administration console
The record is the product
Every learning-significant event — a quiz answer, a simulation pass, a mentor’s logbook signature — becomes an xAPI statement in a tenant-scoped learning record store. Every export, dashboard and model is a view over that one spine, which is why the compliance reports and the client dashboards can never disagree with each other.
Standards at every seam
xAPI everywhere and cmi5 at every launch boundary. LTI 1.3 Advantage in both directions — run external tools inside Academy, or embed Academy modules inside a corporate client’s incumbent LMS. Per-client statement forwarding puts your client’s training data in your client’s own data lake, live.
Engines are replaceable
Anything that merely renders content sits behind a standards seam and can be replaced module by module, with no data migration and no tenant noticing. The layers that matter — identity and authorisation, the compliance domain, the record spine, the AI subsystem and the experience layer — are ours and stay ours.
Governed AI
A per-tenant policy object controls model allow-lists, residency routing, per-role access, rate limits and spend caps. Every call is logged with tenant, purpose, model, tokens and cost. Marks carry provenance — deterministic, AI with model and prompt version, or human — and AI never auto-fails a learner: integrity flags route to a moderator queue, and a default 10% of AI marks are audit-sampled by a person.
Accessibility as specification
WCAG 2.2 AA is the conformance target across web, mobile and admin surfaces. Assessed learner accommodations — extra time, alternative formats, reader and scribe support, captioning — apply automatically across web, mobile, exams and immersive modes, and the accommodation evidence feeds the B-BBEE disability line.
Built for South African conditions
Offline is a first-class citizen, not a degraded mode: downloadable packs, delta sync on updated-at and is-deleted, background flush for progress, assessments, logbook entries, practical observations and simulation telemetry. Load-shedding and a depot with no fibre are design inputs.
Regime polymorphism
Four accreditation regimes. One tenant. No forks.
A transitioning provider is not running one regime. It is running new occupational qualifications, legacy unit-standard programmes in teach-out, a higher-education partnership and a short-course book — simultaneously. Most systems model one of those and make the other three somebody’s spreadsheet.
| Regime | Structural template | What the platform enforces |
|---|---|---|
| Occupational | Qualification (SAQA ID, NQF level, credits, curriculum code, EISA specification) → curriculum component (knowledge / practical / workplace) → module → outcome → assessment criterion | Component-mapped portfolio of evidence; workplace logbook hours; EISA pipeline and its clocks; accreditation scope |
| Legacy unit standard | Qualification → unit standard (SAQA US ID, credits, level) → specific outcome → assessment criterion | Teach-out window enforcement — enrolment blocked past last-enrolment dates — and the File 30 export path |
| Higher-education partner | Partner institution (CHE / DHET registration) → partner programme (NQF 5–10, credits, electives) → partner module | Delivery, formative assessment where delegated, and a defined academic-record export to the accrediting institution. The partner owns statutory reporting. |
| Non-credit | Lightweight course template for short courses, masterclasses and CPD | Certificates without statutory export; feeds the commerce catalogue |
A practical consequence worth stating plainly: legacy-era SAQA IDs in your catalogue need their registration and teach-out status verified during onboarding, so each programme lands on the right regime template. That verification is part of implementation, and we will do it with you before the first learner is enrolled.
White-label and reach
Your brand is the product. The platform is invisible.
Per tenant
- Design tokens — the entire visual system is driven from one variable block, so a re-brand is configuration
- Custom domains with automated TLS
- Email and messaging identities
- Certificate templates and per-client report skins
- Immersive lobby branding
- Reseller mode, in which the platform never appears
Per corporate client
- Sub-tenant branding and its own dashboards
- A filtered live learning-record feed into the client’s own data lake
- Single sign-on federation into the client’s identity provider
- Academy modules embedded inside the client’s incumbent LMS over LTI 1.3
- Optional dedicated compute pools and dedicated database clusters where procurement demands physical separation
Mobile — stated honestly
Roadmap · v1.1 The offline-first learner and assessor super-app is specified and its server-side contracts — delta sync, idempotent batch submission, offline queues for logbook entries, practical observations and simulation telemetry — are built and tested today. The native applications themselves are not yet shipped. We will not show you a screenshot of an app you cannot install.
What they are being built to beat is specific. The incumbent South African mobile learning apps in this market are English-only, carry no compliance layer at all, have no assessor mode, no workplace logbook, no portfolio of evidence and no offline claim. Our targets are five South African languages with speech — isiZulu, isiXhosa, Afrikaans, Sepedi and Sesotho — genuine offline-first operation with conflict-safe sync, QR-based practical observation with signature capture, mentor-countersigned logbooks, and device reach down to older handsets rather than only recent ones. Those are targets we are engineering against, and we will report against them when the apps ship.