Automating Ukko’s receipts accounting with Machine Learning

Advian’s Senior Data Scientist Matti Karppanen streamlined the accounting of Ukko’s customers’ receipts using Machine Learning. The automatisation resulted in time and cost savings, as well as freed a team of accountants for other essential tasks.

Ukko is a SaaS company that offers accounting, payroll, and invoicing services to individual entrepreneurs, such as taxi drivers, barbers, and photographers. They had a great need for tedious manual labor in accounting every receipt sent by their customers – one-by-one. Hundreds of thousands of receipts tied a team of accountants, which was costly and non-scalable as Ukko’s business continues to grow.

The process of accounting receipts had to be automated in order to save in time and costs, and to better allocate resources.


We were able to automate a lot of manual accounting that is both high volume but also fairly easy for artificial intelligence to learn the rules. Advian helped us to in matter of couple of months turn a manual process into automated and thereby save us significant costs”, said Aleksi Simola, CEO of UkkoPro.


To automate an accounting process with clear rules and criteria isn’t that simple. Matti deployed the automation with a Machine Learning approach using AWS SageMaker, NLP, and deep learning. In practice, the text of every receipt is now scanned, interpreted, and automatically accounted if the level of confidence is high. For instance, a receipt from a gas station can immediately be recognized and accounted in the transportation costs account without manual intervention. Matti’s solution resulted in automating a large majority of the receipt handling, making Ukko’s operations much more scalable.


"I was able to deploy a machine learning solution with periodic retraining on AWS using a stack of SageMaker, Lambda, Step Functions, and REST. I enjoyed the active cooperation with Ukko and the opportunities to give presentations about my work", says Matti Karppanen, Senior Data Scientist.



Read next:

The Best Tools (And Some Practices) for Machine Learning Read next:

This time in Advian's blog we will take a deep dive into technical, when Matti Karppanen writes about the best tools and practices for machine learning. Matti is a former professional poker player and a top-notch data scientist, whose specialties lie in linear algebra, neural networks and deep learning.

👉 Read the article here.

Matti blog


Topics: News, Advanced Analytics, Machine Learning



Subscribe to blog

Recipes of unfair competitive advantage for modern BI-leaders.

Get the latest blog posts directly to your inbox


Valikoituja näytteitä tiedoistamme ja taidoistamme