Team Jazz Hands pitches ScotiaLite, a low-bandwidth banking app for seniors, rural users, and newcomers in Canada. It strips away complex UI, adds a 24/7 voice assistant called ScotiaAssist, and uses on-device machine learning to learn what each user needs most. ScotiaLearn offers optional tutorials with mock transactions and incentives.
An AI-powered virtual voice assistant (ScotiaAssist) available 24/7.
A/B testing, multi-armed bandit (context-specific), reinforcement learning, Markov decision process (with neural network), and deep deterministic policy gradient; on-device ML and speech-recognition/NLP pipeline are also noted.
Chapters
Hi everyone, this is Team Jazz Hands and we're proud to introduce skillshare Lite, your alternative online banking mode with easy navigation, minimum Internet requirement and 24. 7 virtual voice assistant. Examining the problem space, we find three leading causes to the digital divide within the online banking sector of the Canadian society. First, while there are continuous increases in Internet and smartphone uses by seniors, they're often confused or uncomfortable navigating those complicated multi step features. Secondly, a significant difference in Internet speed is observed in urban versus rural areas, resulting in rural users facing difficulty accessing complex pages or apps. Lastly, there exists digital banking hesitancy mostly due to minimal digital financial literacy. People are concerned about online security, the lack of face to face communication, whether transactions are overly technical and potential language barriers.
The solution ScotiaLite with more seniors, students and individuals without consistent Wi Fi using online banking, they often find current apps too overwhelming and inaccessible. Scotialite focuses on only listing the essential tabs that match the day to day needs of seniors. An interactive 24.7ai powered virtual voice assistant is available should users prefer audio communication instead of textual searching? Easily adjustable text sizes are available for personal needs. With the pandemic pushing more services online, it further highlights how essential Internet access is for rural areas. While urban areas in Canada have 100% high speed Internet availability rate, rural areas only have 87%. Hence, ScotiaLite is Internet friendly with minimal advertisement and visualizations. It prioritizes frequently visited tabs to allow quicker access. The mode tracks users behaviors and offers timely reminders or directed page landings. Considering a large user base being new users and newcomers to Canada, ScotiaLite offers two beneficial one being ScotiaLearn, an optional digital financial literacy tutorial that includes mock transactions and offers incentives. The other is targeting newcomers who may not be fluent in English. The mode allows language switch ups as well as multilinguistic helper guides for navigation.
So what is the methodology behind all these features? Skillsharelight uses reinforcement learning to power a unique user experience that gives the right information at the right time. Backend naive ABM testing is done to find out which tabs are more frequently visited. Contact specific multi armed bandit help to collect user specific information and optimize user experience by customizing service tab orders. For example, if a user visits credit cards more often it will be listed near the top of the page. Markov decision process uses neural network to predict next and future steps of users. Along with deep deterministic policy gradient. Scotia Lite learns user behavior patterns. For example, if a user always views check in account on Fridays in future Fridays when the user opens Scotia Lite, a direct option to view will be prompted. In addition, ScotiaLite leverages speech recognition and natural language processing by launching the Scotia Assist Virtual Voice assistant feature. This 24.7feature gives prompt and clear information to users such as seniors who prefer audio over navigating textual data. The three steps can be broken down into audio signal detection, word detection and text transformation, and command analysis and feedback output. Moreover, instead of traditionally sending collected data to the cloud, these machine learning tools can now be directly implemented on the device, which is much quicker and utilizes much less data.
When the user first signs into the app, they're given the option to try out Scotia like mode. This mode has a much simpler UI and shows only the essentials like account balance and distinct synced buttons for sending transfers, paying bills and learning more about the app through Scotia Learn. There's also a digital voice assistant that can help you navigate the app. If you prefer audio over visuals, for example, you can ask the assistant to show transactions from the previous month. Also the ability to access settings where you can adjust the different text size for accessibility reasons, as well as going to the Settings tab to access the language settings. This is an example of a translated page in French and then you can also go to Scotia Learn and that is where you'd able to be incentivized to learn more about financial literacy as well as different services that Scotiabank offers. And some examples for incentives could be like Scotia Scene Points. So this is an example of a module that Scotia Learn could implement to incentivize especially marginalized communities or individuals like students or seniors to continue to learn. Scotia Learn will empower users to utilize their digital tools by reducing user friction using this gamified approach with possible scene incentives and increase accessibility to financial services, allowing customers to learn more about services and different financial concepts. Finally, it'll also cover financial literacy and career readiness, which again reduces the digital divide. Light means a more inclusive society. Thank you.
