Making the right decisions.

How/Why we started our AI journey.

written by Iva Gornishka

All companies and products have their background story. Their mission. The reason why they’re doing what they’re doing. Something that guides their efforts. For us, it’s all about helping HR and payroll professionals give their employees the best possible service.

And you know what’s awesome about having a mission? That it’s our driving force. The light on our path. The answer to all of our questions and struggles. It’s our 42.

Creating a roadmap? Deciding on an appropriate solution? Evaluating the results? Easy-peasy. As long as you ask yourself the right questions and follow your mission. As someone wise (or maybe simply had too much time for social media) once asked:

“What can you do today to bring you one step closer to your goal?”

It’s easier said than done and we don’t always get it quite right the first time around, but that is exactly the question we try to ask ourselves every time a decision comes our way.

But let’s see how this all worked out for our AI journey, and more specifically for one of our most recent projects.

Innovation

First, let’s take a brief moment to look at the big picture. As you probably already know, we operate in an industry that still depends much on manual work. We help HR and payroll professionals by automating their daily tasks. If you have read some of our blog posts, you can tell we are very passionate about what we do and how we do it. We know very well that there is only so much we can do simply by automation and following established processes, so we constantly strive to innovate, not only within our product but also in our organization — we always find new ways to look at and solve problems, because we believe this is what will revolutionise the experience we give our clients and the one they give to their employees.

That is how we started as a cloud company, making our customers’ work accessible at any time by anyone and also how we started our microservice transformation, and that is why we follow our version of the Spotify model. Working in an agile environment without managers allows a constant flow of great ideas at Nmbrs, and we are not afraid to experiment with them and see if they help us grow and take another step towards our long-term goals. Does it help us be more scalable? Does it help us go global? And most important — does it help our customers deliver the best service possible to their employees? Then we roll it out!

And this is exactly how we started our AI journey…

The Vision

In a field where automation was unheard of until just a few years ago, buzzwords like AI cause more confusion than the invention of the light bulb did a while ago. That’s the reason why all of our AI-driven features ended up having a whole cute personality around them. We might tell you more about this story another time but at least now you won’t be surprised when we call them “Numbees”.

We have made it our squad mission to be the pioneers in the field, to show payroll professionals that data science is here to make their life easier and help them spend more time on the human touch and expertise they can add.

This all made deciding on a vision quite easy. We had three possible directions to go into for our future AI-related applications. First, we could improve our internal process or their bottlenecks, e.g. analyze the incoming support tickets, improve the search engine of our knowledge base, improve the scheduling of automated payroll processes… But we wanted to help our customers first, and not make our own lives easier, didn’t we? While we constantly strive to help our squads make data-driven decisions, our own AI efforts are focused on improving our user experience first.

Second, we could also explore the possibilities of what we now call “Funky AI” — wouldn’t it be super cool if you could talk to your payroll app and ask it to show you your payslip?! Yes, it is awesome. But we doubt our customers would have the time and energy left to play with their payroll toys after a day of manually inspecting wage tax declarations. But we will get there one day, for sure!

Third, the only reasonable way we could go was to create features that truly help our customers with their day to day tasks. And that is what we vouched to do until we get our self-driving payroll platform.

The Roadmap

Now let’s be honest here. While most of our technical squads have their hands full with customer wishes, and cool new dream-features to implement, our data scientists are still a bit “misunderstood” and our wish list is… let’s say quite empty. But we love that because it gives us the freedom to come up with ideas ourselves!

Well, we didn’t do it completely alone, of course. Otherwise, we’d end up with people sitting all day long reproducing random scientific papers and applying fancy models on completely pointless payroll-related datasets just because we could. What we did instead was to talk to our customers and get to know them.

In these conversations, we learned a lot about the manual tasks payroll professionals perform daily, the data they sometimes need to copy-n-paste four times, the settings that commonly cause problems. While some of the feedback ended up on the backlogs of other squads, a lot of it taught us what are the type of AI-driven features our customers would appreciate. This inspired us for a few projects to come.

So all great, right?! We have an idea of what our customers might need, our mission is to make their daily tasks as simple as possible, so we should be able to easily prioritize projects now! However, we admit it is not always as simple as that in real life. Our system and database were not designed with data science and AI applications in mind, so the actions and feedback we store are commonly insufficient or simply not suitable at the moment to solve more complex problems.

But you know what? We don’t need to solve it all at once. Yes, we have a mission, but also a vision to guide us and an agile culture! So there we go — while we constantly strive to improve our system, adjust the actions we log and the information we receive from our customers to better learn from their behaviour in the future, we can simultaneously start small — analyze few settings at a time, validate small parts of the puzzle, and keep making small steps towards the end goal.

Making the right decisions

So, what can we do today to bring us one step closer to our goal?

We can start small, with just one experiment and see where it brings us.

Continue looking at problems differently and adjust our course accordingly. But most important — do not forget what our goal is, and keep reminding ourselves that everything we do, we do it to provide a great service to our customers and their employees.

How do you make decisions in your team/company? We’d like to know more about the struggles and achievements you go through and how that connects to the product you create.

If you saw yourself in how we do things at Nmbrs, we’re hiring! Apply here.