Skip to main content
The Ontology dashboard — also called the Content Console — is the central hub for everything that powers your AI assistant’s ability to answer questions accurately and consistently. It organizes your medical and informational content into structured knowledge libraries, classifies asset types, maps topics to response templates, and surfaces coverage gaps so you can keep your AI’s knowledge base healthy, current, and comprehensive. Think of it as the editorial control room for your conversational AI.

​
KBase Index Summary

The KBase Index Summary gives you a high-level audit of every content library currently indexed and available to your AI workflows. Each library is identified by an index key (for example, idx:ncx23-medical) and displays three core metrics side by side.

Treatments

The total number of distinct treatment entries in this index. A higher count means the AI has more specific clinical pathways and procedure descriptions to draw from when answering patient queries.

Actions

The number of actionable workflow triggers — such as booking steps, referral prompts, or follow-up instructions — linked to content in this index. Actions connect informational content to operational outcomes.

Response Blocks

The total number of pre-structured reply units available for the AI to assemble into a final response. Response blocks are the building blocks your assistant uses to compose answers.
If the treatment count in an index looks lower than expected, it may mean recent clinical content hasn’t been ingested yet. Check your content ingestion pipeline and re-index after adding new material.
Use the KBase Index Summary to confirm that:
  • All relevant indices are actively loaded (not stale or unsynced)
  • The response block count is proportional to the breadth of topics you need to cover
  • Action counts reflect the full range of workflow triggers your AI should be able to initiate
Multiple indices can be active simultaneously. Your AI workflows query across all loaded indices, so maintaining accurate content in each one directly affects response quality.

​
Format Classification

The Format Classification section breaks down your total indexed content by asset type. This tells you not just how much content you have, but what shape it’s in — which is important because different asset types serve different roles in the AI’s response pipeline.
Knowledge Blocks are the core reusable content units in your knowledge base. Each block captures a self-contained piece of information — a clinical definition, a step-by-step procedure, a FAQ answer, or a policy explanation. They are designed to be retrieved and embedded directly into AI responses.A healthy knowledge base has a large proportion of well-formed Knowledge Blocks. If this count is low relative to your topic coverage goals, prioritize authoring new blocks before adding other asset types.

​
Topical Coverage (KBase Topical Count)

The Topical Coverage section presents visual donut charts that show how your content is distributed across medical topics and specialties. Each chart segment represents a topic area (such as HPV Vaccine, women_care, or Appointment Locations), and its size is proportional to the number of content pieces indexed under that topic. This view answers a critical question: does your knowledge base have enough content to handle the topics your users actually ask about?

Identify content-dense topics

Topics with a large segment are well-covered. Your AI is likely to return high-quality, specific responses for queries in these areas. Use CX Ops trending topics data to confirm that user demand aligns with where your content is concentrated.

Identify content gaps

Topics with a very small segment — or missing entirely — are coverage gaps. If users are asking questions in these areas, your AI will either return a generic response or escalate to a human agent. Prioritize content creation for these gaps.

​
Acting on topical gaps

1

Cross-reference with Trending Topics

Open the CX Ops dashboard and review your Trending Topics list. Any intent that appears frequently in user conversations but has thin representation in the donut chart is a high-priority gap to fill.
2

Create targeted Knowledge Blocks

Author new Knowledge Blocks specifically for the under-represented topic. Aim for multiple blocks per topic — covering definitions, common questions, procedures, and follow-up care — so the AI has sufficient material to compose nuanced responses.
3

Re-index and verify

After adding content, trigger a re-index of the relevant KBase library and return to the Topical Coverage view to confirm the topic’s segment has grown. Then monitor LLM Retrieval Quality in CX Ops to verify that response scores improve for queries in that topic.

​
Template Usage (Topical Template Count)

The Topical Template Count section shows which response templates are actively mapped to specific conversation topics. Templates define the structure and tone of the AI’s reply for a given intent — for example, the AppointmentBook template governs how the assistant presents booking confirmation details, while AppointmentFollowUp shapes how it handles post-appointment check-ins.
A topic with zero template mappings will receive a generic or fallback response, even if Knowledge Blocks exist for it. Template mapping is what converts raw knowledge into a well-structured, on-brand reply.

​
How to use this section effectively

  • Confirm coverage for high-traffic topics — any topic that appears frequently in your Trending Topics list should have at least one active template mapped to it.
  • Avoid template overloading — assigning too many templates to a single topic can cause the AI to select inconsistently. Keep template assignments specific and intentional.
  • Review after content updates — when you add new Knowledge Blocks for a topic, check that an appropriate template is mapped to handle the expected user intent for those blocks.
Removing a template mapping for an active topic will immediately degrade response quality for that intent. Before unmapping a template, verify that another template covers the same topic or that the topic’s traffic volume is negligible.
Use template usage data alongside the KBase Topical Count to build a two-dimensional view of your content health: topics need both sufficient Knowledge Blocks (coverage) and an active template mapping (structure) to deliver a high-quality AI response.