Loomi: the artificial intelligence assistant

As an AI personal assistant, it’s essential that Loomi can be trusted to take care of user data with the highest level of information security. Through Digital Catapult’s Machine Intelligence Garage, the Loomi team participated in an Ethics Committee deep dive that would inform the product’s development and build, and to make sure that the platform is demonstrably ethical and secure. Loomi now also has transparency built into its AI algorithms, so that users can understand at a glance why something is being shown to them.

More than just another voice search engine

Unlike most existing AI helpers, Loomi isn’t just another voice search engine. As an AI personal assistant platform, it automates the tasks that a real personal assistant would carry out, making a high level of personalised service available, and helping to manage information better.
Loomi connects with the individual’s user data (calendars, diaries and so on) via APIs and extracts intelligence to use and self-train. In the organisation environment, Loomi enables on-premise secure aggregation of internal data, build of custom ontologies and ultimately more accurate personalisation of data and tasks for internal users.

The benefits of Digital Catapult’s Ethics Committee ‘deep dive’

The Loomi team joined an early cohort of Machine Intelligence Garage, recognising Digital Catapult as offering the best programmes in terms of networking with other startups, building relationships, exposure to industry and academia, as well as access to different technology resources and research bodies. Because Loomi is a complex and highly technical product, a focus on security, how to handle user data, and providing transparency around how the product works was required – evidencing integrity to be able to earn user trust.

The initial focus was on functionality and being able to reflect the Loomi core values within the products and the business as a whole. Being part of the Ethics Committee deep dive proved to be a significant benefit, enabling Loomi to go into unprecedented detail, and enabling the exploration of a wide range of relevant aspects of product development, including meeting the needs of different types of users, establishing relevant policies, training, communications and marketing strategies.

As well as learning from industry and academic research, the deep dive enabled Loomi to focus dedicated time and expertise on this all-important area of the business. This facilitated questioning, reflection and identifying priorities more clearly, and resulted in new actions bring added to the product roadmap (such as relabelling functionality to make it more user- intuitive to the user, and finding different ways to evaluate sources of news content).

Turning learning into positive action

Loomi’s launch was pushed back to accommodate the deep dive and its outcomes, giving the team more time for product development and addressing the new actions which had emerged. Now there was a structure for implementing and managing ethics, and this output would be used to influence and differentiate the way Loomi was to be marketed.

The Loomi team has continued to interact and network with others on the Machine Intelligence Garage cohort, and recently led a successful Digital Catapult workshop to share learnings on AI ethics and algorithm transparency.

“One of the key things for us is the ability to build trust. Being part of the Ethics Committee deep dive was an imperative for us, as ethics forms a core part of the essential value of our product…The deep dive was a serious time commitment, so it’s not right for every startup – but if this is likely to be a differentiator for your business, I’d recommend it.” – Al Ramich, Founder & CEO
Loomi