Earlier this year we released Contextual Entity Resolution (CER), a post-transcription layer that corrects speech-to-text errors using your business context. CER made transcripts more accurate. Entity Knowledge Graphs take that a step further: they turn those corrected entities into structured, searchable knowledge about your business.
Every customer conversation is full of names, products, locations, and concepts that matter. A caller mentions a specific product line. An agent references a regional office. A customer brings up a competitor by name. Individually, these are data points. Across hundreds or thousands of conversations, they become patterns.
Until now, those patterns lived in the transcripts and nowhere else. If you wanted to know which products come up most in escalations, or which locations generate the most billing questions, you would need someone to manually review calls and piece it together. That does not scale.
After CER corrects the transcript, Compass now identifies the important entities in each conversation and tracks them over time. When the same entity appears across multiple conversations, it gets promoted to a project-level catalog, a living record of the people, places, and concepts your customers talk about most.
Three things make this useful:
Automatic catalog building. Entities that appear across multiple conversations are promoted into a shared catalog. No one has to manually tag calls or build keyword lists. The catalog grows organically from your actual conversation data.
Alias resolution. The same entity often shows up spelled differently across calls. A customer name gets transcribed three different ways. A product has a nickname agents use interchangeably with the official name. The system detects these variants and collapses them into a single canonical entry, so your data is not fragmented across alternate spellings.
Relationship mapping. Compass tracks which entities appear together. If a specific product and a specific complaint consistently show up in the same conversations, that connection is recorded. Over time, this builds a relationship map that reveals patterns you would never spot by reading individual transcripts.
Most conversation analytics tools can tell you what happened in a single call. Few can tell you what keeps happening across calls. Entity Knowledge Graphs bridge that gap.
For operations leaders, this means visibility into recurring themes without building manual reports. Which products drive the most support volume? Which customer segments mention competitors most often? Which locations generate the most complex calls? The answers emerge from your data automatically.
For CX and Sales teams, it adds context that was previously invisible. When a pattern connects a product issue to a specific customer segment, that is actionable. When a competitor keeps appearing alongside a specific objection, that is intelligence you can use.
The system is fully customizable per project. Teams can adjust how frequently an entity needs to appear before it is cataloged, how aggressively aliases are merged, and how many relationships are tracked per entity.
Entity Knowledge Graphs are live for all Compass clients with CER enabled. Like CER, they run automatically with no setup or configuration required.
