NLU for Dialogue is a Unique Machine Learning Problem Consider speech recognition. This is a machine learning problem that also requires a lot of data. This data can be annotated relatively simply by a person listening to a recording and transcribing what is said.  With machine translation, there are many different ways to annotate data […]

APPROACH 1 DIY, from scratch Some companies opt to build their own DIY voice assistant in-house, from scratch.  Bank of America’s “Erica” Bank of America did just that with their AI-driven virtual financial assistant: Erica. This completely self-owned system was created by a development team of 100 people in 2017. It took the team nearly […]

PolyAI CEO & Co-Founder Nikola Mrkšić spoke to Daniel Whitenack and Chris Benson last week on the Practical AI podcast. The topic of the podcast is next-gen voice assistants, and Nikola takes Daniel and Chris on a deep dive into conversational AI, discussing the next generation of voice assistants that will be capable of handling […]

In December 2020, the team at IBM published a blog post and technical paper in which they benchmarked the performance of a new-and-improved Watson Assistant against other conversational AI platforms from Google, Microsoft and Rasa. Naturally, we were very curious.  As it happens, some of the training datasets used in the benchmark test were the […]

Language is full of characters and words that sound frustratingly similar. Sorry, was that M or N? Did you say three, free, or tree? In normal conversation amongst friends, this is easily overcome but customers are less forgiving when they have an issue with a product or service. Before customers even get the opportunity to […]

Let’s face it. Understanding your customers isn’t always easy. They tell stories, go off track and they often can’t think of a good way to explain an issue to you. And those are just the customers that speak the same language as you. Even for humans, it can be really difficult to understand people with […]

In a recent evaluation by Salesforce Research, PolyAI’s ConveRT model performed top across a range of metrics, while using a fraction of the computational resources. Salesforce’s recent paper, Probing Task-Oriented Dialogue Representation from Language Models, compared ConveRT to other pre-trained models, evaluating their ability to encapsulate conversational knowledge in application to Conversational AI tasks. ConveRT […]

We’re thrilled to announce our recently published paper on the PolyAI ConVEx framework. Our new technique, ConVEx (Conversational Value Extractor), is the most accurate value extractor on the market. It requires significantly less data than previous best systems, which means that PolyAI can create virtual assistants faster and better than anyone else, across any customer […]

At PolyAI, our conversational agents are powered, in part, by machine learning models that detect the intent behind what a user says. For example, in a banking environment, if a customer says “When did you send me my new card?”, the models will detect that you’re enquiring about card arrivals and the agent will route […]