How Rydoo got smarter: launching Smart Audit

Have you ever experienced a time when you have so many ideas that you just don’t know which you want to tackle first? That’s how I felt like for the better part of a year while the team was building Rydoo Smart Audit, our AI-powered expense monitoring assistant. And what a journey that was… let me tell you all about it.  

I’ve been a product manager at Rydoo since 2018, and my main focus for the last few years has been on compliance, expense rules and approval flows. My first major project was to revamp our expense rules, a task that took nearly a year to complete but had a great impact on the way our customers manage their expenses. Approval flows followed, with the goal of making them even more flexible and adaptable to the different structures of our customers’ organisations.

There was already so much our software could do, but with AI, a new world of possibilities was at our disposal.

Both these projects originated from the feedback we got from our customers, who often brought questions, feedback and suggestions to our Customer Success team as to how our product could be improved. Being a user-centric organisation, we focus on the satisfaction of our customers, so we gather and analyse all their feedback so we can work on solutions that further improve Rydoo’s expense management solution. And those calls were the little spark that ignited the fire to start building Rydoo Smart Audit. 

Being the main project manager for expense rules and approval flows, I’m often involved in calls with prospects or existing customers to understand their main pain points, the issues they’re facing and how Rydoo can target those challenges for them and make expense management a seamless process. Suffice to say that, over the last few years, I’ve learned a lot.  

Then, something new and exciting joined the chat: Artificial Intelligence. There were already so many things our software could do, but now, with the rise of AI technologies, a new world of possibilities was at our disposal. Soon enough, a lot of ideas started to form in the back of our minds.  

We started looking at all that feedback once again, and, for every call I had, I brainstormed with our Chief Product Officer, Seb, about ways we could use AI to solve their issues. Soon enough, all those brainstorming sections turned into something very real, and a new project started to take shape. And so, in around May 2023, the work on what would become Smart Audit started.  

We already had a lot of feedback from customers and prospects, so to kickstart the project, we aggregated all of these insights and, together with our legal team, scheduled further meetings so we could interview more customers and gather even more information from them.  

The response rate to this request was surprisingly high, and we had a lot of our customers, including some of our biggest enterprise organisations, that immediately responded, showing their enthusiasm to collaborate with us on this project. We proceeded to interview them and were surprised to understand that most of them openly shared the actual pains they faced on a daily basis. Some of these conversations led us to discuss a sensitive topic for most organisations: expense fraud.  

We soon realised that it’s still very painful for finance teams to handle this issue, especially considering how many declared that their highest achiever employees are often the ones committing fraud and the legal implications this can have on the organisation. It became quite obvious that fraud was a big pain point, and it needed to be tackled as soon as possible.  

We officially kicked off the project with a design sprint week with our team, which included myself as a project manager, designers and engineers. For a full week, we fully immersed ourselves in this project. We closed our Slack, turned off email and spent days building prototypes. 

We soon realised that it’s still very painful for finance teams to handle fraud.

By the end of the week, we got back to our customers and had our engineers show these first few ideas and gather all their feedback to start the development stage.

At this point, we encountered a challenge, our lack of in-house experience in AI development. The team debated outsourcing or hiring options, but while discussing this with the team, we realised that our team of engineers was already following up and learning a lot about AI, and they were enthusiastic to get started and further develop their capabilities.

The decision led to a learning curve, as we all had to learn a lot along the way, but the team quickly figured everything out and, in the end, it was worth it, showing us that our team is even more capable than we’d ever imagined. They adapted, experimented, learned a lot and tested different models to build the best possible solution for Rydoo. And the best was yet to come.  

With the team ready to kickstart development, we faced a major issue: where to start. We had gathered a lot of valuable insights that generated dozens of ideas, but we had to make a decision. But what should we prioritise? We officially kicked off with AI keyword detection, as it was a feature that no other expense management software had back then.  

AI keyword detection allows the system to detect out-of-policy items, handwritten receipts and more at an unprecedented level. This saves users significant amounts of time, as they don’t have to manually review all of their receipts, and prevents that some of those small details that the naked eye can often miss to not slip through the cracks.  

All the conversations we had with customers have provided us with use cases for features we want to add in the future. 

Once the first iteration was ready, we shared it with our early-access customers, and their feedback was invaluable. At this point, the software could already translate receipts as well, and they found this to be particularly useful, as they often had employees flying across the world. This will help uncover issues that would have otherwise gone unnoticed.  

We had customers stating that they immediately discovered some suspicious activities. A customer even recalled an occasion when someone submitted a claim for Cuban cigars as a business gift, which the teams wouldn’t have realised if it weren’t for Smart Audit. But because Rydoo flagged it, the customer rewrote the organisation’s expense policy to make it clearer for employees, and easier to handle for approvers and finance. 

When presenting Smart Audit to prospects, their reactions were overwhelmingly positive as well. Many were impressed with the capabilities, noting that they hadn’t seen anything like it in other expense tools. It provides them with more transparency to detect out-of-policy items or receipts, and now they can make their policies also more strict. And all these conversations have provided us with other use cases for features we want to add in the future. 

One organisation, for instance, stated that their team often tried to submit claims for IT purchases, which was not allowed as they had proper channels for it. It took the team a lot of time to review and reject the expenses just to have employees try to submit them again. This is something the right AI tool can easily tackle.  

Once we started discussing Smart Audit with early access customers, conversations started to prepare for the launch, and that meant aligning with the marketing team to make it as successful as the feature itself.  

Having our Product Marketing Manager, Esteban, deeply involved in this project from the start was essential. He had previously worked at a company that used Rydoo as a customer, so he also had that end-user experience feedback and insights to share along the way. The rest of the marketing team also got involved in different stages of the project so we could all brainstorm ideas to make the launch as unique and impactful as it happened with Rydoo Cards in 2023.

After a lot of consideration and discussions, we decided to announce Smart Audit to the world in May. And while we could have waited to build on more features to enhance it even further, we wanted to provide our customers and prospects with the powerful functionalities we already had available, as they could already provide great value to those who use it.  

Whenever I describe Smart Audit to someone, I always say what we’ve shared with everyone so far is just the beginning of what this module can do. Our team keeps bustling with ideas that we believe can further enhance the power of our AI expense monitor in ways that will make it seamless for organisations to ensure compliance, without compromising on their productivity.  

We have big plans for Smart Audit, including transaction receipt detection and more AI keyword checks. We’re also working on Per Diem and mileage checks, as these are some of the easiest ways to commit fraud, and even combining these checks with auto-approvals. This will all drastically reduce the time finance teams and approvers spend on reviewing expenses. 

We also want to have advanced behavioural pattern detection, as it can be a sign that some malicious and fraudulent actions are occurring, and it can often be missed when relying on manual checks.  

As far as other AI capabilities go for Rydoo, we’re exploring AI-driven policy suggestions, a solution that will generate an ideal expense policy based on a few questions and integrate it within Rydoo. This is especially important for smaller organisations that often don’t know where to start with their policies. Approval assistance features are also on the horizon, where instead of rejecting an expense and having to write the reason, an assistant will write this message based on the context provided by the claim. There’s a lot we can do, and we learn more every day. 

We have big plans for Smart Audit, including transaction receipt detection and more AI keyword checks.

Launching Smart Audit marks a significant milestone for Rydoo, as we’ve now entered the AI expense management game. It’s a testament to our commitment to bring our customers the most innovative solutions based on their feedback and experience with the system. We’re excited about what the future holds, and I can’t wait to continue working with the team to make Rydoo even smarter.  

Stay tuned for the next chapters!