What is it?
A 4-month ongoing project as part of my co-op term at RBC, Intelligent OCR is a proof of concept designed to test the Optical Character Recognition capabilities of the Google Vision ML Kit and prove whether or not it will work for the purposes of automating the tedious task of inputting information into a written application.
How can we automate the current manual process?
The use case of this project aimed to tackle the issue of the automatic classification of documents for mortgage applications. The current process involves manual effort to open documents from clients and classify them manually. For the business, this POC will streamline the process and increase efficiency.
Outcome
This project spanned the duration of my co-op term at the RBC Technology & Operations Innovation team from January to April 2019. The project is currently in development - any future changes to the project may or may not be related to the project below.
Manual Verification is a Thing of the Past
To automate the process of scanning and classifying documents for a Mortgage Application.
Currently at RBC, the process of classifying physical documents is very laborious and time consuming – with employees manually verifying each document individually.
The current process at RBC requires designated employees to scan and classify each document.
What’s the challenge?
On the business front, there is an enormous time delay from verifying documents manually.
What are some user problems?
Applicants would send in documents in poor quality, or they would send in the wrong form altogether.
These errors would not be caught until they reached RBC’s Processing Center, by which the Advisors would have to wait again for the client to send a new document.
Project Objectives
Integrate on-mobile-device solution into backbone screens for auto population, starting with the mortgage application flow.
Utilize on-device mobile tech to capture key data points from 3 file types (T4, NoA, Paystub).
How do we help?
We reframed the problem into two statements:
1. How can we increase time efficiency of sending and receiving sensitive documents?
2. How can we increase the security and privacy of secure client information?
Value Proposition for RBC - With each action, users are eased into the complicated mortgage process. While the application is intended for advisor use only, the applicant can still experience the process with a walkthrough – increasing their ease and trust in the advisor.
Documentation Process Research
As a co-op on my team, I was given the freedom to design screens. To inform my decisions and process, I did a lot of research into current mortgage application processes, in general and pre-existing at RBC. On the business perspective, I researched current pre-approval processes and required documentation for mortgages. To inform my design decisions, I researched the user journey framework and flow of a mortgage application process, as well as how that process would be improved and what it might look like with the automated process in place. Insights I discovered during my research is that:
1. Customers want a transparent, easy-to-follow status update on their applications so they are fully informed when decision timelines will be completed
2. There are a plethora of tools available for RBC advisors to use, most of which are difficult to use and include lots of frustrations from user research.
For the Client
For the Business
This would make the mortgage application process more transparent and enjoyable as they are able to view each and every single step of the process.
This would accelerate the time it currently takes to manually process documents, saving time and minimizing cost. It would also reduce human error while extracting information.
Advisor’s Panel for Easy Client Management
The Advisor’s panel helps Advisors keep track of the status of active and complete applications. To display the OCR technology, the Advisor is also able to scan a document to autocomplete a profile.
Onboarding New Technology
The concept of Optical Character Recognition is not new - as evidenced by its daily use on applications like Google Translate or integrated into device cameras on new devices. However, for a user who may not be used to using this feature, those who have none-limited knowledge of how to use document scanners will be able to learn how to use the tool efficiently.
OCR Process
Visually, it was important to have an indication that the OCR is scanning so that users aren’t confused by a screen that could appear to be stuck.
Classification
The OCR technology automatically classifies the scan according to the document the applicant is scanning.
Verification and Editing
Applicants will be able to verify their information that the OCR has pulled. They can also manually edit the fields after the human verification component. For security purposes, there is also an “edited” view and “view original”.
Accountability to Reduce Human Error
To remove the possibility of scanning and verification errors, both the client and advisor has the obligation to sign off on the documentation. Once the documentation process is complete, a Call to Action allows the possibility of scanning more than one document to complete whole application.
My Reflections as a Co-op Student
On a personal level, this was my first time handing off designs to developers in a professional capacity, and I was able to work through challenges of not knowing the process to being able to work with the developer team and compile assets and documentation for them.
This project (and co-op) gave me the opportunity to work in an interdisciplinary team in a business. I experienced a lot of different career paths within this co-op - product management, graphic design, UI/UX design, technology and development. The phrase “jack of all trades, master of none” used to scar me, but I found that it does not apply. I came out with the realization that being able to know and educate myself about everything makes me a better designer, because then I am able to synthesize all these different parts of a project and use my understanding to inform and be an advocate for my design decisions.
Accolades
Along with other RBC intern projects, my work was nominated to be shown in front of key stakeholders and subsequently, I was awarded Winter Co-op Student of the Term for 2019 out of 300 students.