Group X Expenses NLP
Razorfish formed GroupX a team focused on innovating, bringing 9 Creative and UX team members with one technologist to create demos, assist in pitch work and proof of concepts.
One such project was revamping the expense entry project.
Expense entry is a horrible experience.
one needs to manage receipts, and enter in all the right fields, but what if it could be better?
The design team took a look at integrating Optical Character Recognition (OCR, for scanning receipts into meaningful information) and voice interaction.
At a time when Siri was just coming to market and before Alexa and Google home were not available, voice interaction was in it's infancy.
Our approach
As the sole engineering supporting a team of experience designers we focused not on the visual design which changed weekly, but on demonstrating technology aspects that were reaches at the time.
First a speech to text library was used to provide text of the conversation.
From there we performed linguistic analysis to look at breaking down words to their types (noun, verb, adjectives, etc) and then use that to parse information out of the speech
We grabbed as much data out of this to prepopulate a form of all the required data.
Legacy lab experimentation (older than 2020)
Lab descriptions and details currently not available online, please reach out to discuss.
