Share this post:
How many times have you looked at your bank statement, noticed a line item like “WITHDRAW / POS 0217 2013 975246 MARSHALLS #1199” and racked your brain trying to figure out what you bought?
You likely had to dig through your wallet for the paper receipt (if you even kept it) or rummage through your inbox for the electronic copy – but no longer.
Now, when banks incorporate the receipt management capabilities of Sensibill into their mobile apps, customers can see purchase details and more in the palms of their hands. Sensibill integrates with banks’ existing mobile apps and web interfaces for an easier way to keep track of purchases, expenses and budgeting.
How Sensibill works
When a customer takes a picture of a receipt with a smartphone, the picture is sent to the Sensibill service based in the IBM Cloud. Sensibill then extracts the data and creates an enhanced receipt; a smart receipt.
Sensibill’s smart receipt system can remind you about return and warranty information, as well as pull information about the merchant, including a contact number and address.
Sensibill uses a proprietary algorithm to process any receipt with a high level of accuracy. Since there are approximately 150 data points on receipts, Sensibill counts on an R&D team of experts whose single mission is to extract receipt data and classify it.
Machine learning: An essential component
The R&D team specializes in machine learning techniques – deep learning, specifically – to read and understand receipts. This means identifying items, the categories items belong to, the merchant, the total and so on.
The idea is that as the machine continues to “learn” and become more complex, it will be able to recognize the item “Levi’s 501 jeans”, and categorize it as clothing or men’s apparel. Understanding the true identity of an item means that the machine can provide deeper, enriched insights.
Proprietary technology and advanced algorithms enable Sensibill to take full advantage of machines to perform work at a scale that is just not possible for humans.
Sensibill’s “machine” can also learn to understand historical and habitual purchases. With that information, users can track spending and budget, or banks can use it to present customers with appropriate advice or offers.
As the Sensibill system continues to learn and add layers to its hierarchical neural network, it will be able to discern increasingly complex information based on customers purchase data. Beyond identifying items, it will understand the meanings attached to common spending patterns until can perform basic inductive reasoning.
Ultimately, the Sensibill system will be able to predict a customer’s needs in real time based on changes in behavior, then alert customers to offers they need and want, maybe before they even know they need or want them. This capability will augment bank services profoundly.
How banks benefit
Sensibill developed its solution to work with banks concerned about fintechs disrupting the industry and eroding their customer bases. The solution has been built from the ground, with security in mind, and offers a robust experience.
Banks often do not employ subject matter expertise outside their core offerings, so rather than try to build something like this on their own, they can partner with Sensibill. Incorporating the receipt management solution into its service greatly improves a bank’s ability to serve and retain customers.
Sensibill credits its ability to work well with banks to its receipt management solution running on IBM Cloud. Banks trust IBM and its solid expertise in the financial services industry.
Read about other IBM clients who built their success on the IBM Cloud here.