Every Company Speaks Its Own Language
Your organization has terminology that means something specific to your team:
- “Hot lead” — maybe that means responded within 24 hours with budget confirmed
- “Code red” — maybe that means a client escalation requiring same-day response
- “Whale” — maybe that means a deal over $100K
- “The Johnson playbook” — maybe that means the multi-touch enterprise sequence that worked with Johnson & Associates
This vocabulary is tribal knowledge. New hires take months to learn it. AI systems that don’t understand it produce generic, context-free outputs.
What Ontology Mapping Does
Ontology mapping encodes your organizational vocabulary into structured data that AI agents can use:
- Terms are defined with precise meanings in your context
- Hierarchies are established (a “whale” is a type of “enterprise lead” is a type of “lead”)
- Relationships are mapped (a “code red” triggers the “escalation playbook” which involves the “client success” team)
- Thresholds are quantified (“hot lead” = engagement score > 80 + response within 24 hours)
Why It Matters
Without ontology mapping, AI agents work from generic definitions. With it, they work from your definitions.
Without: “Follow up with hot leads” → The agent uses a generic notion of “hot” and contacts anyone who opened an email.
With: “Follow up with hot leads” → The agent identifies contacts who responded within 24 hours with confirmed budget, checks the knowledge graph for relationship context, and uses the appropriate sequence from your Skills Registry.
Building Your Ontology
Ontology mapping isn’t a one-time setup. It evolves as your business evolves:
- Start with existing documentation — SOPs, sales playbooks, onboarding docs already contain much of your vocabulary
- Capture usage — As teams use the system, new terms and meanings emerge naturally
- Refine continuously — When the system misinterprets a term, that’s a signal to update the ontology
- Version control — Like your Skills Registry, your ontology is versioned so changes are tracked and reversible