Build a Great Nex

What makes a great Nex

A great Nex is not a long Nex. It is the smallest set of concepts, activities, and citations that get a learner to durable mastery, with evidence a regulator could reconstruct. Three constraints separate the great from the average:

• Outcomes over hours. A Nex is judged by what learners can do afterward, not by minutes consumed.
• Mastery over completion. Confidence and competency by concept, not a single “passed” flag.
• Evidence over polish. Every claim cites a source. Every assessment leaves a defensible trail.

The best Nex is the one a regulator could audit, a manager could cite, and a learner could pass on demand, six months after they took it.

Start with the outcome

Before opening the builder, write three to five sentences a learner should be able to say out loud after the Nex. Tie each to a node in your knowledge hierarchy. If you can’t map an outcome to a concept, the outcome is too vague or the hierarchy is missing a node. Fix that first.

• Use action verbs: identify, escalate, decline, document, redirect. Avoid understand and be aware of; they aren’t observable.
• Each outcome maps to a knowledge byte. Knowledge bytes drive the assessment plan downstream.
• If two outcomes share a concept, collapse them. Redundant outcomes waste learner time and dilute mastery signal.

Source from your truth

Point The Brain at the policies, SOPs, recorded SME calls, decision memos, and prior incident reports that already define your truth. Never seed a Nex with a generic web crawl. The quality of the source corpus is the upper bound on the quality of the Nex.

• Prefer first-party documents. They are versioned, owned, and citable.
• Tag every source by domain, owner, and effective date so the regulatory monitor can flag drift.
• If a topic has no first-party source, that is a content gap, not a Nex problem. Close the gap upstream.

Outline the spine, then expand

A great Nex has a spine of 6 to 12 modules, each anchored to exactly one concept. Use the AI Course Builder to stub the spine from your outcomes, review it with a real SME for thirty minutes, and only then expand each module into lessons and activities.

• One concept per module. If a module covers two, split it.
• Module length: 8 to 18 minutes of learner time. Anything longer is a sign you’re hiding two concepts.
• Open every module with a one-sentence outcome and close with a mastery checkpoint that proves it.

Mix activities, never lecture for long

The fastest predictor of a forgettable Nex is a wall of slides. Cap any single explanatory block at 8 to 12 minutes, then put the learner in motion. Pick activity types from the library that match the concept, not the medium you’re used to.

• Decisions: branching scenarios and decision trees, not multiple choice.
• Procedures: hotspot, drag-to-order, and process simulations.
• Conversations: AI voice role-play with tone and pacing scoring.
• Recall: spaced retrieval cards, not the same quiz repeated.

When in doubt, the best activity is the one closest to the real moment the learner will face on the job.

Cite every claim

Citations are the difference between a Nex you ship to regulators and a Nex you have to defend. Make every paragraph point to a source policy or document, and surface the citation in the learner UI, not just the audit log.

• Every AI-drafted paragraph carries a citation by default. Reviewers can swap citations inline; never strip them.
• Assessments record the question, answer, source, and concept. That is your evidence trail.
• When a source policy changes, the regulatory monitor flags the Nex with a severity score so the right paragraphs get a refresh, not the whole course.

Calibrate difficulty for the cohort

A great Nex doesn’t make every learner sit through every concept. Use mastery checkpoints to retire bytes a learner already knows and to extend practice on the bytes they don’t.

• Set the target mastery threshold per concept based on risk. A 70% bar on a courtesy topic is fine. A 92% bar on AML escalation is not negotiable.
• Use adaptive paths so high performers finish faster and strugglers get more reps, not the same Nex twice.
• For new hires, anchor the first checkpoint inside the first 15 minutes. Early signal beats end-of-Nex regret.

Ship, measure, fix the weakest

The first version of a Nex is a hypothesis. Ship it to a small cohort, watch the mastery heatmap for a week, and fix the bottom decile of bytes before opening it to the org. That loop is faster than a quarter of authoring polish.

• Look at the byte-level heatmap, not the module score. Modules hide the failing concepts.
• A byte under 65% mastery is the signal to rewrite the explainer or replace the activity, not the question.
• Use AI agents to draft the fix from your sources, then review. Authoring time on the second pass should be minutes, not days.

A 7-day rollout

The teams shipping the most influential Nexes follow roughly this rhythm. It collapses what used to take a quarter into a week.

• Day 1. Write outcomes. Map them to nodes. Confirm with one SME.
• Day 2. Ingest sources. Tag by owner and effective date. Spot-check citations.
• Day 3. Stub the spine in the builder. Cut modules ruthlessly.
• Day 4. Expand modules. Replace any block over 12 minutes with an activity.
• Day 5. SME review. Swap citations inline, not paragraphs.
• Day 6. Pilot to a 20-person cohort. Watch the heatmap.
• Day 7. Fix the bottom decile of bytes. Publish org-wide.

By day eight, you stop asking who completed what and start asking what your workforce knows. That is the real measure of a great Nex.