hu:toma AI

To provide the tools and platform to build, share, and leverage deep learning chatbots

Founded 2016
1-15 employees
  • Information Systems
  • Headquarters address
    London

    How did it start?

    Our CEO's personal project to teach his son to speak English (without an Italian accent). Maurizio whilst working for Amazon in the US leading their Machine Learning Development Team for the Amazon Go decided to put his skills to use. He wondered if he could build an AI that would fit inside a toy that his son played with and in-turn help him with his English. A Kickstarter project followed, forming hu:toma with his brother the company was successively scouted and accepted onto the Startupbootcamp IoT and Data program in Barcelona.

    Vision & Growth

    Startupbootcamp helped hu:toma shape the product into a commercial prospect, together we saw the potential application within all customer-facing domains. Being Hackers at heart Maurizio and Andrea wanted to leave the individual creation of bots up to the collective knowledge of the wider-world and instead provide the tools and platform to build, share and leverage deep learning chatbots.

    We're building a platform that allows Third-party and corporate developers alike to train an AI using only a text file, no need to code OR map the conversation tree which can quickly become complicated. This can then be posted to the "Bot Store" where it can be combined with other AI's - this allows for monetisation and open-source.

    Team

    CEO - Ex Amazon & Microsoft Machine Learning Tech Lead (Phd in ML)
    VP Business Dev- Former Google & Uber
    Growth Team - Former Startupbootcamp EIR
    Engineering Team - Predominantly ex Microsoft & Amazon

    Interview Process

    We're a startup so nothing is too formal. This will comprise of both a technical & personal interview with both the CEO and one or more of the Engineering team. One or both of which will be over Skype.

    Tech stack

    Python backend and uses a number of deep learning techniques to train a conversational AI. Heavy use of of recurring neural networks. Exposed via REST APIs and a graphical console. API: Java Front end: PHP.