The curriculum is a living system
Static catalogs rot. Ours is designed to learn from licensed practitioner briefings, permissioned learner artifacts, and operating metrics that show where the map is wrong.
Catalogs rot
A course recorded in January is folklore by the following January. Models version, APIs rename, context windows grow, and patterns that were clever last year are antipatterns now. The shelf life of AI content is measured in months, and the economics of recorded courses push the other way: the more expensive the production, the longer it has to stay on the shelf to pay back. The result is an industry quietly teaching last year's craft with this year's marketing.
We didn't want to fight that gravity with heroic re-recording schedules. We wanted a curriculum that expects to change — built from small, cheap, frequently replaced units instead of big, polished, permanent ones. Small units also mean honest authorship: a fifteen-minute briefing carries one expert's name and one claim to maintain, not a production team's reputation to defend.
The Daily Pulse
The intended feeding mechanism is the Daily Pulse: a short practitioner briefing attached to the graph nodes it serves. A briefing on tool-contract discipline lands on the contracts nodes; a field report from a healthcare deployment lands on the regulated-systems nodes. The mapping and publishing surfaces are built. The live feed activates only when licensed briefings from vetted practitioners exist.
Each published briefing will carry a freshness window. When the underlying practice shifts — a model version changes the pattern, an API deprecates — the node can flag stale and the briefing can be replaced rather than footnoted. The goal is a map that is honest about its own decay.
Cohort N teaches cohort N+1
The second feed is designed to come from learners themselves. Cohorts produce artifacts, and permissioned artifacts can become reviewed examples attached to their nodes while recurring failure modes inform the rubrics. When many learners misdesign the same tool schema, that is not merely a set of individual mistakes — it is a signal about the teaching. The next cohort should meet a better example at exactly that node.
Nothing is reused without review and permission, and identities are stripped by default. But the principle stands: a cohort's struggles become the next cohort's guardrails.
When twelve learners make the same mistake, it's not twelve mistakes. It's a curriculum bug.
The practitioner roster as a content engine
The third feed will come from the practitioner roster now being recruited. Vetted operators are intended to contribute field reports, demos, and lab notes from current work under signed publishing terms. As coverage grows across regulated industries, engineering, product, and executive practice, the curriculum can inherit that range without pretending a content studio is the frontier.
The design bet is concrete: practitioners teaching from current work have less reason to invent hypotheticals. The useful examples arrive with an operating trace behind them.
What we measure
A living system needs telemetry or it is just churn. The platform is built to watch operational measures rather than vanity: time-to-unstick per node, first-pass rate per artifact rubric, briefing freshness once practitioner publishing activates, and gate-review outcomes at each level. Watch time appears nowhere on that list. The curriculum is working when people get unblocked and ship work that passes review.
When a metric sags, the fix is specific: rewrite a briefing, add an example, tighten a rubric. Small units, small fixes.
What stays constant
For all the churn, the architecture underneath is stable: the graph, the rubrics, the gates, the portfolio you build. That's deliberate. The units of content are disposable; the structure of capability is not. A learner should never wake up to find their progress invalidated because the industry shipped a new model version — the node they cleared still means what it meant. What changes is the material that gets you there, and the examples the node points to.
Living system, stable skeleton. That's the whole design.