Designing Your Knowledge Hierarchy

Why hierarchy is the bedrock

Every Nex, mastery score, evidence trail, and regulatory monitor in Nexera maps to a node in your hierarchy. If the hierarchy is muddled, every downstream artifact inherits the mud. If it’s clean, the rest of the platform stops fighting you.

Think of the hierarchy as a schema for your workforce knowledge. Authors, regulators, dashboards, and AI agents all read from the same nodes. That shared spine is what makes evidence reproducible and analytics comparable across teams.

The six levels at a glance

• Field. The industry. Financial Services, Healthcare, Industrial, Tech.
• Subject. The thing regulators name. AML, Patient Privacy, Workplace Safety.
• Domain. A coherent slice of the subject. CDD, EDD, Sanctions Screening under AML.
• Topic. A grouping a course can teach in one sitting. PEP screening, beneficial ownership.
• Concept. A single idea. Enhanced review triggers, sanctions list ownership.
• Knowledge byte. The atomic claim a learner must master. “Triggers for enhanced review on a PEP relationship include X, Y, and Z.”

Knowledge bytes are where mastery is measured. Everything above is organization.

Name nodes regulators use

The fastest way to break the hierarchy is to invent your own vocabulary. Use the names regulators, auditors, and your policy library already use. Your hierarchy should read like a regulatory index, not an internal taxonomy.

• Borrow subject names from regulations: AML, GDPR, HIPAA, EU AI Act.
• Borrow domain names from your policy library, not your org chart. Owners change, domains shouldn’t.
• Avoid product or vendor names at the concept level. They date the hierarchy.

Map content to nodes, not the other way around

The discipline that separates clean hierarchies from messy ones: nodes come first. New content gets attached to existing nodes. New nodes only get added when a regulator, policy, or SOP introduces something genuinely new.

• Every Nex maps to a topic. Every module maps to a concept. Every assessment item maps to a byte.
• If you can’t place a piece of content in the hierarchy, the content is unscoped. Fix the scope before shipping.
• Tag every source with the highest node it cleanly covers. Avoid forcing one document onto a single byte.

Versioning and ownership

Every node should have one owner and a version. When the regulation moves, the regulatory monitor flags the node, the owner triages, and the courses that teach it queue for review. Without ownership, drift is silent. With it, drift is a notification.

• Owner = a named person, not a team. Teams diffuse, names respond.
• Version every node when its meaning changes, not when its wording changes.
• Keep a changelog at the concept level. Auditors love it. Authors use it.

When to split, when to merge

• Split a concept when two distinct mastery signals are getting averaged into one. The cohort is good at one and bad at the other and you can’t see it.
• Merge two concepts when nobody can write a question that distinguishes them. They were the same idea, renamed.
• Retire a node when no policy, SOP, or regulation references it for two cycles. Dead nodes clutter dashboards and confuse new authors.

A great hierarchy is small enough to fit in your head, big enough to cover what regulators ask, and stable enough that new courses just plug in.