Four frameworks govern how competency is defined, how instruction is structured, how evidence is recorded, and what has to be true before a learner is called competent. This page documents each of them, and the reasoning behind the design.
A ladder implies uniform ascension: one path, fixed sequence, every learner starting at the bottom. A scaffold has multiple entry points at different heights, multiple exits, and platforms that can be occupied independently of one another.
This distinction is the claim a reader has to accept before the rest of the architecture makes sense, and it is the property most easily lost when the model is reduced to a diagram of ascending strands.
Applied to competency architecture it produces three requirements. Entry position is determined by evidence rather than enrollment date, so recognition of prior learning is the ordinary case of entry at a non-zero position rather than an administrative exception. Position is per-competency rather than per-learner, so a learner may hold mastery on one competency, mid-developmental progress on a second, and no evidence at all on a third, simultaneously and without contradiction.
And rate is unconstrained, differing across competencies within a single learner. Speed-to-competency is a property of the learner-competency pair, not of the learner or of the program.
The scaffold property is what makes the architecture compatible with differentiated instruction, adult learners with heterogeneous backgrounds, and prior learning assessment. A ladder model can accommodate none of these without treating them as exceptions — which is why programs built on one spend so much administrative effort on students who do not fit it.
The argument is developed in full, with its grounding in assessment validity theory, in The Dual-Ascent Competency Model™: Structurally Separating Developmental Skill-Building from Workforce-Ready Mastery Demonstration in Competency-Based Education Credentials — published open access under CC BY.
A competency architecture that tracks every competency twice — once on a developmental strand recording skill acquisition, and once on a mastery strand recording demonstrated proficiency — and never permits evidence from one strand to alter the record of the other.
Most competency systems record a single determination per competency: achieved, or not. That record cannot distinguish a learner who passed a knowledge check after a tutorial from one who performed the task under evaluated conditions and was assessed against a rubric by a qualified assessor. Both are plausibly written the same way.
The consequence falls on whoever reads the record. An employer, an accreditor or a regulator infers readiness from a document that does not contain the information the inference requires. This is a structural problem rather than a measurement problem: no improvement in the quality of individual assessments resolves it, because the deficiency is in how evidence is stored and rendered.
Non-fungibility is the load-bearing constraint. Architectures that permit a weighted composite, a best-of-both resolution, or promotion of developmental evidence to mastery status on a threshold have re-merged the constructs. Separation that can be collapsed under any condition is not separation.
The two strands answer different questions and are measured differently. The developmental strand answers whether a learner can perform a skill today, and is scored on a graduated scale because the information of interest is trajectory. A low early score is expected; the shape of the curve is the signal.
The mastery strand answers whether a learner has proven readiness, and is binary. A professional judgment that a standard has been met is not usefully expressed as a percentage. Averaging the two produces a number that answers neither question.
The developmental record is permanent. Remediation adds entries; it never amends or removes prior ones. A record revised to reflect present standing is not a developmental record — it is a mastery record with extra steps, and the architecture has collapsed back to a single layer.
Every unit of instruction moves through three phases — Educate, Demonstrate, Validate — with a gate between each. The cycle is the structure of the lesson, not a feature layered on top of it.
Educate is knowledge acquisition: reading, video, interactive modules, self-checks, adaptive warm-ups. It produces evidence into the developmental record, and its checks must be passed before a learner advances.
Demonstrate is application in context — practice under realistic conditions, on equipment where the discipline has equipment. It also produces developmental evidence. In aviation deployments this phase is named Aviate, under the mark Educate. Aviate. Validate.™; other verticals substitute domain-appropriate terminology while the underlying structure is unchanged.
Validate is competency verification: instructor-scored rubrics, scenario-based assessment, formal performance evaluation. This is the only phase that writes to the mastery record, and it is where a qualified human makes the competency decision.
The gates are the point. Because each phase must produce its own evidence before the next unlocks, a learner cannot read their way past a skill or practice their way past a knowledge gap — and an instructor can see, for any lesson on any day, which phase of the cycle each learner is standing in.
Forty-two standards across eight domains. Every course built on the platform is constructed and audited against them.
Twenty-five years of being measured against other people's standards produces a fairly specific opinion about what a good one looks like. The standards are stated as constraints rather than recommendations: a course that violates one has reintroduced the problem the architecture exists to solve. A school adopting the platform inherits a quality framework without inheriting a third-party subscription.
An integration layer that connects training equipment to the competency record. Where a skill has to be demonstrated on a device, CirroSync captures what the learner actually did and writes it in as structured, timestamped evidence.
The problem is common to every technical discipline. The demonstration happens on the equipment; the record of it happens later, on paper or from memory, and by then the detail is gone. Programs are then asked by regulators and accreditors to evidence competency progression, and what they have is attendance and a signature.
CirroSync is hardware-blind by design. A thin adapter normalises each equipment family into a single evidence schema, and everything downstream of that layer — scoring, routing, reporting — is identical regardless of what produced the data. Equipment profiles are configuration, not compiled software, so supporting an additional device does not require a new release.
Scoring is anchored to the external standard rather than to the device. Each connected device carries a fidelity class that informs how its data is treated and travels with the evidence, so a reviewer years later can see not only what was demonstrated but what it was demonstrated on.
CirroSync does not make the mastery decision. Equipment data is evidence, not judgment. Promotion to the mastery record requires a qualified human. It does not replace instructor authority, does not issue regulatory authorizations, and does not alter an approved program — it observes and records.
Cirrostratus™ is a multi-tenant, white-label competency-based learning platform. The frameworks above are not documentation about it — they are enforced in its data model.
Evidence routing is structural: practice-venue evidence is written to the developmental ledger and evaluated-venue evidence to the mastery ledger, enforced in the data model rather than by configuration, so the rule cannot be circumvented by well-meaning local adjustment. Every entry carries the conditions under which it was generated — instrument, venue, supervision status, and the identity of the evaluator where a professional judgment was involved.
When a competency falls short, the engine traces the gap to the specific external standard element where it occurred, identifies the cognitive level at which it manifests, and selects intervention content from pre-approved, human-authored templates. That pipeline is deterministic, auditable and reproducible.
An optional AI Study Companion is available to learners and is architecturally isolated from it: the two never share decision pathways, and the companion has no influence on mastery determinations, remediation triggering, or credentialing. The platform is AI-assisted where assistance helps a learner, and deterministic everywhere a decision has consequences.
The engine is field-agnostic. It currently carries three verticals — aviation, healthcare, and technical education — each with domain-appropriate terminology at the surface and the same architecture underneath. The aviation vertical is delivered as CSAP, the Cirrostratus™ Supplemental Assessment Program, which layers competency measurement onto training a school already delivers.
If you are evaluating whether this architecture fits your program, the intake form is the fastest way to a useful conversation. It asks what you train, what standard you train against, and what evidence you are being asked to produce. Both paths are reviewed directly by Dr. Lescarbeau.
Or read the underlying scholarship first. The Dual-Ascent Competency Model™ →
