DigiForm

• 

  Process workflow

• 

  Sample Form to upload

Inspiration

As an old aged person, I am not able to cope with digitalization of forms. I still prefer to fill paper application than online application. Will be happier, if I don’t need to visit bank to do it But, I need to get products suggestions that suites me.

What it does

DigiForm does digitization of paper forms using various solutions, including, Responsive Web Apps, Blob Storage, Micro Services, Microsoft Vision API, Cosmos DB, FFDC APIs, AI / ML, hosted in Azure Cloud Environment. This will enable self-service for an old aged person to fill his application in paper yet get the advantage of filling forms digitally. Once the user uploads the form, it goes through workflow of micro services where the form data are captured. The captured information will be sent to bank through FFDC. AI/ML helps the user to predict possible interest rates and suited product. FFDC APIs would help him know the loan status in real time.

How we built it

We built the following components and orchestrated them to achieve the objective • Responsive website • Collect basic information • Take Picture / Upload image • Verity uploaded form data • Know loan Status and product suggestions • Blob Storage • Stores the uploaded images • Micro Services • Create zones • Consumes Microsoft Vision API • Collect information and stores it in Cosmos DB • Microsoft Vision API • To fetch handwritten information from the image • Cosmos DB • Stores telemetric information • Stores form values temporarily
• FFDC APIs • CRM Imports to Import loans • Loan status to retrieve loan status • Rate search to get the product details • AI / ML • To Predict preferred Products and rates

Challenges we ran into

• As the photo of the form is to be taken by the customers who are old aged, there were chances of getting edges of image not captured properly, making the zoning failed. So we made changes in algorithm to capture data using relative spacing. • Once the data is captured, before routing it to bank officer, we had to come up with some logic to route it to the right person who can serve the customer best. We used geographical location to identify whom it has to be assigned.

Accomplishments that we're proud of

• Reduce the workflow turnaround time of paper form applications by 40% • Consumed 7 endpoints of 3 FFDC APIs. • ML - Trained using over 3 Lakh records to get predicative loan rates and best product selection

What we learned

• There is a huge amount of customer still in market, who wish to fill their application in paper format • There are not much software available to cater these customer group

What's next for DigiForm

• Convert Any physical Forms • Can be used to convert any physical files adding a classifier service • Can be used to collect supporting documents • Read data from supporting documents • Information can be captured from supporting documents submitted • Validate filled-in information using data from supporting documents • Data captured by Supporting Document can be used to validate the data filled-in by the user

Built With

• ai/ml
• api
• blob
• ffdc
• micro-services
• responsive
• storage
• vision-api
• webapp