105234Re:Infer - Deep Learning Conversational AutomationClosedFeasibility StudiesInnovate UK"Increasing complexity of business practices, changing market and customer requirements and intensifying competition means that the need for process automation is becoming increasing critical. Regardless of sector or size of operation, significant volumes of conversational data are generated daily through the interaction between a business and its customers as well as through internal communications. Despite the potential value of this data to improve its customer experience, operational activity and competitiveness, the ability to truly analyse the content of complex interactions generated by the large volume of phone calls, emails, on-website chat generated by internal and external customer communication is far from optimal with a current reliance on manual interpretation to perform this activity. With significant variations in content, the ability to automatically extract both the meaning of a communication as well as the intent is a highly complex task which is not possible through current call analytics products or AI based solutions which on rely specific metadata, key phrases or sentiment analysis. Since 2016, advancements in Natural Language Processing have proven the ability to extract and respond automatically to structured data from natural language. However, despite this potential, current approaches have focussed on understanding well written text and typically require significant volumes of structured annotated data for the systems to be trained. These technologies are therefore not designed to be deployed in commercial application where there is little annotated training data and where communication can vary significantly in grammar, spelling, content containing business specific terminology With potential across multiple sectors (initial target being large Financial and Insurance companies as well as Ecommerce and Telecomms), and based on the use of proprietary Deep Learning technology and a unique system design, Re:Infer offers the first automated conversational intent recognition system capable of interpreting unstructured communications data from any form of communication converting this from natural language to structured data -- without the need for software expertise or additional coding input. With market need validated and with support through Innovate UK, an 18-month programme of research is required to deliver a prototype which will be validated in a user trial. If successful in development, the solution has the potential to disrupt the way customer communication is currently analysed providing greater insight into a customer's behaviour, offering improved operational intelligence and process efficiencies through automation as well as increased sales through better direct marketing efforts."