Next-level automation: building smarter banking in FreeAgent

Dec 13, 2019 · 4 min read
A glimpse into the future of banking in FreeAgent

Categorising your bank transactions in FreeAgent is a key part of our users’ small business accounting, as it helps them track their incomings and outgoings. We call this process ‘explaining bank transactions’. With a business to run and many transactions happening every day, however, it can be a real challenge for small business owners to keep on top of these explanations.

Recognising how tricky this process can be, FreeAgent can learn over time from how you’ve explained past transactions and suggest explanations for future ones. We call this feature ‘Guess’, and we’ve seen that it can save our users both time and energy when it comes to staying on top of their daily admin.

Building on our existing Guess functionality, we’re now investing in machine learning methods and focused workflows to try to make Guess even better — and we’d love for you to help us get there! Read on to find out more about the project and how you can sign up for early access.

Doing a bit of ‘Guess’ work

While Guess can help with much of the heavy lifting when it comes to explaining transactions, we recognise that it can still come with some frustrations.

To get going, Guess needs historical explanations to learn from, which means that it can take a while for new users to start to see its benefits.

There is also a lack of fine-grained control over how and where Guess is applied. For example, there’s currently no way to teach Guess based on the changes you make to its suggested explanations. The only options are to keep Guess switched on — and fix it manually when it gets an explanation wrong — or to just turn Guess off completely.

Using research to give us direction

Over the past few months, we’ve been running a series of research sessions with the aim of better understanding how Guess is currently used, as well as identifying opportunities to improve and expand automation across the whole of the banking area.

Through these sessions, we have identified three key areas to focus on:

  • Increasing visibility and understanding of how Guess works
  • Giving users greater control over the intricacies of the Guess feature
  • Saving users time and letting them focus more on the big picture

To implement these, we’ve come up with a whole heap of ideas for ways we can improve this feature. We’ve been hard at work getting some of the first elements ready to test.

Taking the first steps

A preview of the new and improved Guess

Our team have used machine learning to create an exciting new way of automatically providing explanations for bank transactions. This new method can do far more than our current Guess algorithm — and utilises more than just the contents of your FreeAgent account.

We‘ve’ taken the categories our users use most frequently to explain their bank transactions and then trained an algorithm to identify common patterns.

For example, it turns out that most of our users explain Starbucks transactions as ‘Accommodation and Meals’, so if we identify a Starbucks transaction, our new algorithm will automatically suggest an explanation of ‘Accommodation and Meals’.

At first, the new algorithm will try to nail these ‘day-to-day’ business transactions like train tickets and takeaway coffees, so that our users can focus on explaining the trickier stuff instead. At the same time, we’ll also be updating the UI in this area, to try to better highlight when there are transactions awaiting approval.

Sign up for early access!

Would you like to experience this new functionality before anyone else? Through our early access program, we’re taking the approach of incremental change and learning more with every step, with continuous updates based on the feedback we receive.

By taking part in our early access program you’ll be guiding our future development in automation in banking. You’ll also be one of a select group of people who will see some of the exciting changes first.

If you’d like to be part of the early access program, please sign up here. Thank you!

— — — — — — — — — — — — — — — —-

Craig Clarke is a Senior Product Manager in the Banking team at FreeAgent - the UK-based online accounting software made specifically for freelancers, small business owners and their accountants.

More From Medium

Also tagged Machine Learning

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade