Tier 01
Individual
"What can I do, what should I learn next, and where could I go?" A personal mirror, coach and navigator — capability growth that belongs to the person, not just the org chart.
How it works
From a job description
to a training plan.
One pipeline ingests the work you need done and returns an evidence-cited plan to build the capability to do it — gap-analysed against the people you already have.
The flow
JD → skills → gap → training.
Ingest the job description
Paste a real JD — or pick a seeded one. SMART treats it as the definition of the work that needs doing, not a static document to file away.
Dissect into required skills
The JD is broken into the skills it actually requires, mapped onto the ontology. Where the work needs a capability the taxonomy doesn't yet cover, SMART mints the novel skill rather than dropping it.
Assemble the ideal, evidence-backed skill set
Each required skill is attached to its graded evidence pack — what actually builds that capability best — producing the target a strong performer in the role would meet.
Gap-analyse against a real person
The ideal set is compared to an incumbent's live skill graph. Gaps are value-weighted so the most consequential shortfalls lead, and coverage is expressed as a single honest percentage.
Emit a hyper-specialised training plan
The output is a prerequisite-ordered, evidence-cited plan with the best-known intervention per step — then delivered and measured through the copilots, with the skill graph updating as learning completes.
Who it's for
Four tiers, one source of truth.
The same capability graph serves four different questions — so an individual's growth and an organisation's strategy are reading from the same data, not two disconnected systems.
Tier 02
Manager
"Who on my team can do this work, and where are our gaps?" Team coverage, task-to-best-person allocation, and the next-best training action per report.
Tier 03
Director
"Is my function fit for what's coming?" Ideal-workforce design, projected gap analysis, and capability tied to the profit levers the function owns.
Tier 04
HR / L&D
"Where is knowledge at risk and is our training working?" Bus-factor and retention, intervention effectiveness over time, and capability analytics across the whole workforce.
The engine
Deterministic where it counts. AI where it helps.
The scheduling maths — gap detection, prerequisite-respecting ordering, skill decay, coverage — is computed by a deterministic, testable engine. The AI is used for rationale and content: assessment narrative, coaching steps, microlearning, the career story. So "adaptive" is real, not a vibe.
A learned, adaptive memory
SMART keeps a longitudinal record of each person's capability and — crucially — which interventions actually moved each skill, and by how much. Over time it learns the best-known intervention per learner, falling back from person to cohort when history is thin.
Grounded, cited AI
Copilots retrieve the governing SOP or evidence with vector embeddings (lexical fallback if no embedding endpoint is reachable) and answer with citations. With no model key configured, every copilot degrades to deterministic, engine-grounded content that is clearly labelled as an offline fallback — a faked response is never mistaken for a live one.