Breeze Through Banking is the vision and purpose of FinGlide. A highly secure and smart voice API that can swiftly integrate with any web or mobile app in a BFSI landscape, offering an instant upgrade and delight in user experience along with a consistent revenue channel for banks and credit unions, especially digital and challenger banks. Boasting 99% accuracy in voice-to-action conversion, FinGlide is built with facial recognition, voice bonding, pseudo phrasing, biometric and OTP features to provide the end user with secure voice-based access to all channel banking features.
Inspiration
The use of voice is simply easier than text. It is more personal, secure, convenient, and less stressful. This inspired us to leverage the great value that voice-based intelligence could bring to the industry, leading to the research and release of FinGlide.
What it does
FinGlide provides app-based and API-based solutions, both on premise and as SAAS that bonds to the end user, converts his/her voice into actionable API calls and delivers text and voice-based results and actions.
How we built it
It was built by a senior research and development team in the Byte Blanket Fintech (BBFT) labs. FinGlide was built as an android and iOS mobile apps using REACT NATIVE technology with a design and structure that allows it to also be used as an API integration to any banking app as well as web-based banking application.
Challenges we ran into
Analyzing and detecting multiple voice modulations for one person and using data science techniques to understand various gestures (smile, straight-faced, sad) and provide advanced authentication options was challenging but we overcame it successfully.
Accomplishments that we're proud of
Demonstrable accuracy in voice-to-action conversion has been at 99% in FinGlide, compared to several popular, leading voice-based agents that come built-in with Android and Apple phones and speaker-based devices. Rolling out FinGlide in additional languages like French, Spanish and Arabic within a few weeks, is our most special achievement till date. Creating innovation around pseudo phrasing (an invention by BBFT, where the actual spoken phrase only makes sense to the user of the app and cannot be decoded through human intervention) is one of our major achievements as it provides great security over privacy when using voice in a public environment.
What we learned
We uncovered new areas during the build of FinGlide, such as bonding voice patterns, handling accurate conversions, designing listeners that blend and understand speech models, among several others.
What's next for FinGlide
We will continue our research into multiple dimensions such as recognition of various other facial and hand gestures to convert into actionable elements, decoding and processing of complex differences in the regional use of languages, biometric and signature analysis research, voice-to-voice mapping etc. The first release of the product, both as an app as well as API, is ready to be integrated with banking apps and we have generated high interest from several leading digital banks. In parallel with our continuing research and development, we would like to look at integration and distribution options and programs available with Finastra.
Log in or sign up for Devpost to join the conversation.