Rossum Reflects

Hard to believe, for us anyway, that it’s been just two years now since we set out on the journey to build Rossum, and automate manual data entry. I think it’s really important to take time at moments like this, and try to sum up where we have been, how we got here, and where we are going – us and the people who have joined us on the way, like our employees, investors, partners, and of course our customers.


Rossum team has doubled. Elis is able to save 97% of manual work. It is used by Fortune 500 companies. A lot of new features coming in the next year. Thanks to all great people that support our mission - to create a world without manual data entry.

What have we really learned?

When we started this journey, we really thought that we were solving a data problem. A technology problem. Today, I would put it another way. I would say that we are solving a very human problem. The problem is that processing invoices, receipts, claims, statements and the gazillion other types of documents our customers have to deal with is frustrating, it’s tedious, it’s stressful, and basically it just sucks a lot.

The fact is that accounting and finance go way beyond pushing some numbers around on a screen. That’s a tiny part of it, but still a huge amount of work.

Organizations don’t process data. People do. Millions of men and women in back offices are doing this work every day, and what we’ve learned is that most of them feel they could be using their time much better. We’ve learned that behind the eyes of a million people doing routine data entry, is a person with the potential to do so much more. Not only the potential, but the desire as well.

Those are our people. They are the ones we are helping. We’ve always seen Rossum’s technology as a way to augment what people can do and make them better at it. It is that, but it’s also simpler than this. It’s also about freeing people to do what they are really good at and really enjoy: thinking, planning, and solving all tough situations that threaten the operation of their company.

How We’ve Grown

While we’ve been learning, we’ve also been growing in numbers. The size of our team has doubled, and we’ve gone from the apprehension of moving into a “big” new office, to the realization that this office (that used to seem huge) is now quite small for our needs.

We’ve also been releasing a new version of Elis, our flagship invoice processing app, nearly every week for more than a year. Imagine, two years ago we were wondering if the important, busy people at huge companies would take us seriously. Now Fortune 500 enterprises are using Elis in their day-to-day operations. When I stop to think about it, I still get a bit of butterflies in my stomach. It’s a big responsibility, but a huge honor to be trusted this way. It makes all of us want to earn that trust with the quality of our work and our conduct.

The Thank You’s

The Rossum Team Investors Early Adopters
Walking into our office every day fills me with pride. I didn’t know I would ever feel this way, but I do. I am so grateful to every one of you for joining us in our journey. Just having someone believe in you, even before you fully believe in yourself, can give you the confidence you need to take on big challenges. I’m continually thankful to our investors for believing in us. Let me tell you something: you’re not really an inventor if nobody has a use for your inventions! Those first people who took a chance on Rossum and tried to use the first versions of what we were building made Elis and everything else a reality, rather than just a theory. You’ve been crucial, and you must know that people like you make our work possible.

Once again, Thank You!

Has Elis helped people?

One of the key milestones for automation of any kind is whether it performs as well or better than a human doing the same work. Well, over the last year, Elis has achieved this milestone, in a manner of speaking.

Specifically, in 66% of cases, Elis has become confident enough to fully automate data entry without any meaningful chance of errors. In 34% of cases, Elis asks for human supervision (because she lacks the confidence to move forward), but in only 4% of cases, does the human supervisor feel that Elis has made a mistake. That means that Elis has achieved a 96% accuracy rate, which is on par with human-level accuracy in data processing.

Many people in our industry have told us that this was simply not going to happen. That 96% was not achievable. So we’re very proud to say we’ve proved them wrong so far.

Of course, that achievement doesn’t mean that much if it doesn’t actually save you work, does it? This is why we have also been focusing on the number of keystrokes and clicks saved by using Elis during data processing. Elis has achieved a 97% savings of keystrokes and other manual actions from analysts and data processing specialists. Wrists around the world should rejoice at this news!

Before Rossum Elis

155,000 keystrokes
8,000 clicks

per FTE a month

4,150 keystrokes
1,450 clicks

per FTE a month

97% manual actions saved

Elis Has a Face

In building Rossum, we’ve focused on human-machine collaboration. That means that there are two equally important aspects of the technology: the Artificial Intelligence (AI) that does the number crunching, and a User Interface (UI) that efficiently and clearly connects the user of the software with that AI so they can work together.

The AI+UI approach is crucial to the success of Elis. We have focused hard on making Elis ergonomic for operators of the software, and we have given Elis the ability to learn from the user’s behavior to adapt to your needs with time.

One of the worst trends in software development is the idea that people should “learn” a software and figure out what it can do. That’s backwards. Software should learn what you need, and figure out how to do it for you.

Adding new skills to Elis? It’s super easy

Elis started as an open platform, and it will remain so. We’ve been approached over the last year by many developers interested in customizing and adding features to Elis. We’ve been working with a lot of them to improve Elis’s extension environment so that people can easily add skills and Elis can get stronger and smarter. Below you can see two extensions in action.

Google translate extension:

Vendor matching extension:

Looking Forward:

Expect at least twice as many skills from Elis in 2019. Some of the new features will be faster automated learning, higher levels of “eyes off” automation, easier configuration and customization, and one cool new feature that will change the way computers work with tables. You’ll see….


Leave a Reply