What More 🎯 Can You Do with Chatbots?

A few posts ago we talked about “analytical data support” for machine learning models in production. Analytical data support for machine learning in production bridges the gap between building a model in the silo with neatly-processed training data and the complex production scenarios where model performance can drop for all kinds of mysterious reasons.

This is not solved in many industry verticals and every company has its own unique problems. When I was researching what kind of chatbot companies are out there during vendor RFP, a company called Qbox caught my eye which specializes in data for chatbot applications.

The company was launched in 2017. Here’s what they do:

The punchline on the top of their landing page says:

Conversational AI testing tool

With QBox™ you can test, understand and fix your chatbot’s behaviour in a matter of minutes.

My first impression was: I am already producing precisions and recalls, what more do I need to test about “my chatbot”?

What is the biggest pain-point in building chatbots?

…the primary challenge… is producing or obtaining…