KBo has always helped you analyze trends, find example conversations, and pull evidence straight from your own data. We are thrilled to announce a significant upgrade to how much reasoning KBo can do in a single ask.

A question with more than one layer to it used to mean breaking it into several separate questions and stitching the answers together yourself. KBo now works through a layered question the way an expert would, pulling together the right combination of techniques and handing back one answer with the reasoning shown.

Think of the earlier version of KBo as a very good line cook. Hand it an order for a single technique, find conversations that mention a word, calculate a percentage, flag a correlation, and it executed that technique cleanly and fast. But hand it an order with several layers to it (what's causing this, where does it show up, and what should we do about it), and it would default to the one or two techniques it knew best and treat that as the whole dish. You got an answer, just not the full one, and you were left to do the next round of digging yourself.

Improved KBo is the same KBo, now with the versatility of a chef who can run every station in the kitchen and combine them onto a single plate. Ask it a layered question, and it reaches for exactly the combination of techniques that question needs, then plates the result as one coherent answer instead of a single ingredient.

Take a complex, multi-faceted question:

"What are the common challenges or frustrations that aren't addressed and result in a transfer or further action?"

Before: KBo may have filtered to calls that mention "transfer" and have an Eddy attached, the friction signal, and reported the overlap: 41% of friction conversations also contained the word "transfer," out of 1,320 matching conversations. That's a keyword and a flag, useful as a rough proxy, but does not answer the whole question.

Today: same question, but KBo combines "ingredients" to deliver a much more comprehensive answer. For example:

Not every answer will follow this precise logic. The combination changes based on what you're actually asking. KBo decides which ingredients a given question calls for and runs them in whatever sequence gets you there, then shows its work.

We encourage you to experiment with the real questions you have, rather than trying to simplify them for KBo. Here are some examples of questions that may not have resulted in a complete answer before, but would fare much better now:

"Why are we seeing more calls about copay assistance this quarter, and is it the same issue everywhere?"

"I want to understand why calls end in negative sentiment and what agents could do better to avoid it."

"What's actually driving repeat calls on the same issue, and which ones are worth fixing first?"

"How many PA calls end unresolved? When the calls are unresolved, what happens to the patient next?"

"Are calls ending in negative sentiment more likely to be repeat calls? What's causing our negative repeat calls?"

KBo's purpose hasn't changed: exploring your data to get you quick, useful, and insightful answers to your most pressing questions about your conversations. What's improved is the reasoning underneath it, and starting October 2, every KBo user has it.

If you have a question that's always needed a few follow-ups to fully answer, try asking it as one layered question instead, and name the specific outcome you want to understand. Let KBo build the answer in one pass. If a response feels shallow, ask it to go deeper on one part rather than starting over, KBo can build on its own previous answer the same way it always could.

Questions about what's possible now? Reach out to your Authenticx team.