Rindegastos, a Chile-based cross-platform application that digitizes the expense reporting process, had a problem. It needed to process receipts from Chile, Argentina and Peru, but being able to read the data from different countries and detect which country the receipt was coming from turned out to be a tall order.
Can The Currency Conundrum Be Overcome?
Without the ability to properly extract the information from the receipts, the company couldn’t solve its currency conundrum. It started evaluating its options, and talking to vendors around the world about what they could offer to unlock the data coming in.
In 2018, CEO Cristian Yoma Schrader started working with Infrrd. From the technology itself to answering questions and quality of service, he says he’s been happy with his choice. It’s a relationship that is leading the company toward its accuracy goals — Rindegastos expects to reach 80% accuracy in the coming months — and helping it offer the kind of service it previously thought was impossible.
“They told us to just send a batch of receipts in for testing, and then we saw the results,” he says. “We saw it actually detected correctly the specific data that was different for every country. It was amazing, actually. That’s when we saw it was possible to sell a solution that AI could adapt for every country. That’s what we needed, and that’s what we achieved with Infrrd.”
A New Product Offering Using Intelligent Expense Processing
Stringent regulations in Latin America demand a lot of information for expenses submitted, in addition to the date, merchandise and category, he says. There’s tax ID, document number, type of document — and Rindegastos makes it so that users don’t have to spend hours scanning images and uploading them or typing all of the information in. The company puts AI in place to do the work for the users.
From the differing uses of decimal points, different currencies and different tax IDs, there was a lot to do, Yoma says. Infrrd built out a proof of concept that made it clear the company was up to the job. “The work we did in that proof of concept with the team at Infrrd was so good, the problems of the business were solved there,” he says. “Infrrd taught us that the data extraction algorithms are flexible. That’s something we didn’t know.”
Read More
Originally published at https://www.infrrd.ai.
