Learn more

Approach

We are developing something that goes beyond dealing with general commands to understanding domain-specific conversations. Our approach represents a powerful paradigm, enabling people to plan journeys and lookup transit information in a natural way. Leveraging extensive deep learning technology, we are researching and developing different core components.

Background

The way in which people interact with transport services and navigation software is obsolete. These kinds of form-based graphical user interface paradigms used by applications today were developed in a time before technology was advanced enough to communicate with people naturally.
To unlock the power of natural language interfaces, a new approach is required. With domain-specific AI, people can accomplish interactions that would be impossible with today's general form-based tools.

  • Natural Language Processing

    Natural Language Processing

     

    Our Natural Language Processing Engine is a new and responsive approach to the unpredictable nature of real transport-related conversations. It can discover user intents from natural language queries through both speech and text. We constantly improve the engine, ensuring that it covers queries from all horizontals of a domain.

  • Deep Learning

    Deep Learning

     

    We utilise a combination of state-of-the-art Deep Learning techniques based on recurrent neural networks and convolutional nets to power our NLP Engine. With the right data, this combination allows robust and accurate semantic analysis of almost any user query in the transport domain.

  • Graph Databases

    Logic Engine

     

    To get the best out of virtual assistant technology, we are developing a logic engine with the goal of handling the interaction between the user and assistant and providing another layer of intelligence on top of our NLP. Our logic engine is dedicated and optimised for multi-query conversations rather than one-line, memoryless commands; it will represent an innovative approach needed for true conversational interfaces.

  • Data Sets

    Data Sets

     

    One of the biggest determinants of the quality of a neural network is the data it is trained on. We are developing large-scale and rich transport and navigation datasets for our learning algorithms which will ensure optimal performance. These datasets are expanded and improved in real time as users interact with the system - the more it is used, the better it will get.

"Leveraging our extensive machine learning technology, we see a great opportunity to create the ultimate transport assistant - one that could actually understand queries like a human"  - Hami Bahraynian, Co-Founder