About Mid-term goals NLU Manifesto Team Join usWe target the most demanding and biggest NLU market - Enterprise contact center phone calls automation. Direct competitors include Google DialogFlow, Microsoft Luis, Rasa, PolyAI and others
Virtual Agents are direct competitors to Human Agents and Enterprise customers do not want to compromise phone call customer service quality - automation solution has to be capable of reaching same KPIs as implied on Human Agents
The core functionality of NLU besides conversation processing is gathering record data. Speech is a new digital user interface that will allow companies to extract insights, personalize communication, optimize interactions, measure and analyze every piece of conversation
Until there’s General Artificial Intelligence human Dialog Owners will always need full control of virtual agent behavior. Even when dialogs reach tens of millions of combinations they should be governed by humans in a scalable way
Great dialog building toolkit and intent discovery model upgrade framework in hands of an experienced Dialog Owner is the perfect combination for optimal pace of conversation design and development
The more experienced the Dialog Owner, the better first version of a dialog is BUT dialog redesign is a constant, iterative process for any conversation
Fast iteration in dialog improvement is the key to successful redesigning. Key part of Dialog Owner’s toolkit are good analytics and issue discovery algorithms are necessary to properly react to errorneous events
Generic ASR engines are great and if you are lucky they will work fine, but most automation cases require a solution that enables building and maintaining custom mini-ASRs to correctly handle particular expressions
First-choice approach to training intent models is Deep Learning. In high-volume contact center processes the ability to adapt models to new edge cases and covering possible widest forms of intent expressing is the key to keep up with human-like KPIs