What's Up with Tech?
Tech Transformation with Evan Kirstel: A podcast exploring the latest trends and innovations in the tech industry, and how businesses can leverage them for growth, diving into the world of B2B, discussing strategies, trends, and sharing insights from industry leaders!
With over three decades in telecom and IT, I've mastered the art of transforming social media into a dynamic platform for audience engagement, community building, and establishing thought leadership. My approach isn't about personal brand promotion but about delivering educational and informative content to cultivate a sustainable, long-term business presence. I am the leading content creator in areas like Enterprise AI, UCaaS, CPaaS, CCaaS, Telecom/5G and more!
What's Up with Tech?
Unlocking AI-Driven Workflows: Rossum's Intelligent Document Processing, Administrative Evolution, and Future Automation Trends with CEO Tomas
Interested in being a guest? Email us at admin@evankirstel.com
Unlock the secrets to the future of work with AI and automation in this illuminating episode featuring Tomas, the visionary CEO of Rossum. Discover how Tomas's journey from AI scientist to co-founding Rossum led to a groundbreaking approach in business-to-business communication through intelligent document processing. Learn how Rossum's cutting-edge, cloud-based solutions drastically reduce manual errors, enhance working capital, and speed up transactional processes. Tomas shares invaluable insights into the broader implications of automation, stressing the importance of a specialized focus and safety in these advanced systems. The conversation unveils the secondary benefits of automation, such as faster shipping and fewer operational mistakes, showcasing its transformative impact on modern business operations.
Witness the evolution of administrative jobs and the shifting landscape of outsourcing as AI becomes more sophisticated. Tomas delves into the transition from rigid, rule-based systems like RPA and ERPs to more adaptable AI solutions that offer cognitive flexibility. Understand how roles in the workplace are set to change, with employees shifting to oversee and train AI systems instead of being replaced. Explore the trend of businesses bringing processes back in-house with increased efficiency and hear Tomas’s practical advice on implementing automation solutions successfully. This episode is a treasure trove of knowledge for anyone interested in the future of work and the dynamic changes AI is bringing to the administrative sector.
More at https://linktr.ee/EvanKirstel
Hey everyone, fascinating chat today as we dive into the future of work and what the future of work means, with AI and automation for administrative work. Really interesting and important topic. Tomas from Rossum, how are you?
Speaker 2:Hello, I'm good Hope you are good as well. Thank you for having me. I'm good Hope you are good as well. Thank you for having me.
Speaker 1:Well, thanks for being here and really intrigued by your team and mission.
Speaker 2:Why don't we start with that? Tell us about yourself and your journey to Rossum over the years. I'm currently acting as a CEO of Rossum, but I started in the tech field. I actually met the other two co-founders in an AI PhD lab, so I started as an AI scientist and then we spotted the opportunity. We felt that we can make the world a little bit better. So we started the company. We started Rossum.
Speaker 1:Fantastic and I'd love to hear about the mission. And you know it's all about automation, which has been around for a while, but super powered by AI in areas like intelligent document processing. But what's the big idea behind Rossum?
Speaker 2:Yeah, look, we keep hearing about automation right Everywhere. If you look at it in more detail, most of the automation is focused on what's happening inside the companies, but what we have realized is that a big portion of the business processes is actually happening in between the companies. Especially, every business is collaborating with some other businesses and that collaboration requires communication. So companies transact with each other, right. So you're ordering something, you're invoicing somebody, you send delivery note, packing list. Every part of the transaction is usually communicated between the businesses, and the way they communicate with each other is usually by our documents, and those documents can have different forms, they can come in different formats and they can come also via different channels.
Speaker 2:So a lot of that is actually happening in between the businesses, and that's what we focus on. We built Rossum to sit in between the businesses and help them with these transactional documents to automate and orchestrate this process, because the fact that every business is slightly different and uses different internal systems results in a fact that they don't understand each other. You generate an invoice and send it to another business, for example, or back English or purchase order, and usually there must be somebody who helps you to process those transactions. And that's just the beginning. So we built Rossum into this cloud-based LLM layer that sits in between the businesses and helps you to communicate with your business partners.
Speaker 1:Wow sounds intriguing business partners. Wow, Sounds intriguing. So you know, this cloud layer, I mean. Is automation here about reducing errors? Is it speeding up tasks like invoicing, reporting? How will it sort of change the paradigm of work with that back office? We call it? Or?
Speaker 2:admin role so it influences pretty much everything. You know when. This is the thing that bothered us. If company A, let's say, invoices company B, it usually takes several days before something that sits in one business system in company A appears in the business systems in company B. So it takes time, it takes a lot of manual work. Because it's manual, it typically contains a lot of mistakes.
Speaker 2:Sometimes the transactional documents are simply like the companies cannot agree whether it's correct the transaction. So sometimes there is an escalation step where you know one company tells the other, hey, we cannot accept this transaction. Plus, still, a lot of this processing requires some internal context that you need to add into that. It's not just about understanding some document. And finally, still, because it's so manual, we don't actually expect everything from those documents. So a lot of insights stay hidden in those unstructured data. So ROSM provides to its customers a solution that can remove all the manual work. That means that you see the transactional data instantly, which is really, really important for the business you know, for your working capital and even like reporting, obviously, because it's less manual.
Speaker 2:It contains less mistakes, but you can also automate things that are other steps that are attached to this document processing right. So imagine that you receive purchase order. That means, because you can do that all faster. That means you can ship faster and you can get your cash faster. So it has impact on working capital as well, plus the insights you know. Companies are missing their their early payment discounts. Companies um pay late payment penalties. Companies very often ship the wrong product because they they made a typo when processing a purchase order. So so the secondary impact of that is actually even bigger than the primary, which is making it less manual.
Speaker 1:Oh, fantastic. Well, less manual we can all get behind. So you mentioned areas like procurement, supply chain very important tasks, but what other sort of everyday tasks in admin work do you think will be most impacted by automation?
Speaker 2:I honestly believe that pretty much all of it. Uh, obviously, different companies focus on different angles, right, and I do believe that this focus matters because if you really want to automate and at the same time that's this is really important we need to automate those processes safely. You know, it's not just about taking some large language model and pretend it works. No, we cannot rely on that. You, you need to have guardrails, you need to, you need to make sure that if you are processing those transactions, it's it's safe and compliant with your operating procedures. So I do believe that focus matters.
Speaker 2:There are different companies focusing on different parts of admin, and actually understanding the job, understanding the task deeply, really matters if you want to automate it. So there are many, many different types of admin work. We, as Rossum, focus mainly on how can we speed up and make processing of the transactional documents as frictionless as possible, but there are other companies that are focusing on different angles. So what I do believe in is that a big portion of the admin work will be automated, but it will not be done by a single company. There are specialists.
Speaker 1:Got it, and so RPA as an approach has been around for quite a while. Tremendous gains. It's amazing how many companies aren't yet leveraging automation. But your vision is more than just automation. What AI opportunities are not being addressed in the RPA world today? Do you think world?
Speaker 2:today. Do you think so? First of all, where I describe distinction between ROSM and RPA, I usually explain it by the difference. Like most of the RPA runs inside the companies, while we are really focusing on being in between the transactional parties, staying in the cloud allows you and really build the whole solution around. That allows you to solve situations that that require both parties to to cooperate. Like you know, when you receive a transaction that that cannot be accepted for whatever reason, and you need to communicate back to the sender, they, they, they upload you a new version, like so. So that's a slightly different task than being hidden behind, like corporate firewall and, you know, do the typical RPA thing in the user interface. So that's the biggest difference. You know, looking at the slightly different problem and approaching it differently, rather than like what's AI, what's not AI. Because ultimately, I do believe that all the solutions and we see it in the RPA space as well will heavily benefit and use AI. So I see AI more as a tool, more as a technology, and then it depends on the vendors how deep they understand the problem they are solving and how how well they can apply it. Um, that's the distinction. Um, however, speaking of ai.
Speaker 2:You know most of those solutions and being at rpa, being at ERPs, was very rule-based, right, and what the new AI revolution brings you is less rule-based approach to automation. But what we see in the market it also brings a lot of problems, right, because all of the hallucinations, all of the you know the fact that it's you don't fully control it, it's not deterministic, is a big problem Because certain processes in the business need to be deterministic. Right, like you don't want your AP department to run on deterministic process because that's why we've all built those standard operating procedures to actually make it as compliant with the company's rules as possible. So it's a scale right. On one scale you have the old-school RPA that is very algorithmic. In the other extreme, you have just a large language model that can say and do whatever it wants, very undeterministic, lack of control, and both have their problems, right, rpa typically quickly breaks.
Speaker 2:Some small thing changes, it breaks. The only AI approach is very flexible but not under control, non-deterministic, and I do believe that the way forward is somewhere in between, to have the cognitive skills of the AI, to have the flexibility of AI to solve the slight changes. But you still want deterministic process because you want to stay in control. You cannot come to CFOs of the world and say I'm going to run non-deterministic reporting for you. That's not going to happen. That's not going to be accepted.
Speaker 1:Fascinating, amazing insight there. So there's lots of talk about automation, replacing jobs. You know, do you think the admin jobs that we're talking about are at risk of disappearing or you know they're going to evolve into something different? I mean case in point here in the US, when I go into my doctor's office, you know, I see obviously doctors and nurses, and then there's, you know, some room with seven people doing paperwork, or you know, healthcare insurance providers and authorizations and government agencies, and it's just, it's just unbelievable the amount of admin required. So, you know, are those jobs going to just disappear in that room in my doctor's office? Are those jobs going?
Speaker 2:to just disappear in that room in my doctor's office. They will evolve heavily. What we are seeing and look, we see this every day right, we are building a solution that is being used by people in the back office to automate a big part of their job. We actually see a really interesting trend. They are really happy about that, you know, especially new generation, like millennials and Gen Z. They are even surprised. You know they don't. They are surprised by some of the tasks that are still done manually. They were born with their phones in their hands, right, so they kind of expect that being solved by the technology. So it's actually an interesting trend that on one side, you have the new generation that doesn't want to do the job and on the other side, you have a technology that is enabling us to really automate the job. Because I spend quite a lot of time thinking about what's going to happen and it's probably normal that these two things happen the demographic change and the mindset change together with the technology evolution. So it's an interesting thing and I do believe that the admin job will evolve heavily towards where you now have 10 people or 100, maybe even 100 later.
Speaker 2:Uh, you will have one person doing it, but that person's focus and skills will shift slightly because in order to automate on such a scale, you will need to give feedback to an AI system.
Speaker 2:You are there basically to to manage the AI system and we are seeing it because the way you, for example, rossum, is building this user interface for AI augmented work. So you are already augmented by AI when you work in Rossum environment, but at the same time, you are giving feedback to an AI by using that environment, and the way how you work in this environment really matters because you can use it in a way that it's easier for the AI to understand what you wanted to do and it can pick it up quickly and automate better. Or you can use that environment in a wrong way and AI is confused. So the job will evolve heavily. There will be less people doing the work, but they will do different work. They will be still doing admin, where AI needs their input, but they will need to do that in a way that AI can pick up that knowledge and can automate on the next try. So you will be acting as a supervisor and trainer for the AI.
Speaker 1:Fascinating. The other trend we've seen over decades is offshoring and outsourcing, especially here in the US. I mean, do you think this new approach will change the way we think about outsourcing or offshoring and even move jobs back to the US in particular, for example? How do you see that playing out?
Speaker 2:We are already seeing this. We see it every day that businesses come and say, hey look, I outsource this process and it's not working for us. For that reason or another, I want to bring it back, but I want to bring it back with 20 times higher efficiency so that it makes financial sense for the business, right? So I don't see this being a trend. I look like a future trend. I think it's here and it's happening.
Speaker 1:Fantastic. That's good news. So typically, when you get into large enterprises, any change projects are notoriously difficult and expensive and take months or years. You know many of them fail, of course. Think about ERP projects that turn disastrous I mean with Rossum out. How quickly can you get started? What is the process for deploying terms of scope and and? Um, you know, do you have to have, uh, sorry, someone like Deloitte or Ernst and Young get involved in, uh, in the transformation? Sorry, I lost my voice.
Speaker 2:No worries, we work with partners around the world that help our clients to not necessarily deploy the solution. That's fairly easy, but you need to map and understand how your future process will look like right and how you will adapt your process for the new generation of technology. So usually the implementation and deployment of the solution takes a couple of months. But what we tend to do already during the procurement slash, sales process really understand the client's need and really start to map out how the whole workflow should look like better, being with ROSM or without ROSM in the new age. You know, Because there are some surprising things that you bump into right, Like just, and I'll give you an extreme. I'll give you an extreme, but it's still reality.
Speaker 2:We used to exchange most of the invoices in a paper form, right, and email. When email started it was a minority channel. So what businesses started to do because they had the whole process built around a step that was like like we scan an invoice and then email started coming in and it was simply easier, instead of introducing a new sub process to print that invoice and scan it again. It was simpler because it was such a small fraction of traffic and you would be surprised how many businesses nowadays receive 60% of their invoices via email, but the process stays the same, so they are printing it and scanning it a few minutes later. So this one is an obvious and a little bit extreme example, but you can find similar examples across the whole process. So it's about really understanding what needs to be achieved, but also what the customer needs, because those enterprises have all their own specific requirements, but they also typically need to understand that it's good to look at the process, like you know, in the 21st century and really rebuild the whole flow.
Speaker 1:Fantastic, so you're in the beautiful Czech Republic in Prague, but if I'm a I don't know healthcare system here in Boston, listening, how do I get started? Can I just go to the website and start testing the product? How do you suggest people get started?
Speaker 2:Yeah, that's one thing that we've always wanted to have at Rossum, you know, a trial where people can go and try it, because you hear a lot of like. You know, there is so much hype around AI and not many companies are willing to go, especially enterprise software companies willing to go, especially enterprise software companies willing to go out there and showcase the product. So with Rossum, you can go to our website and test the product right away. However, if you want to explore some complex automation case which all enterprise customers basically have, complex automation cases I would recommend booking a demo on our website. Enterprise customers basically have complex automation cases. I would recommend booking a demo on our website and explore it with a specialist who can take you on tour, show you what can be achieved. But in my experience, I'm a tech guy so, for example, I rather play with the trial rather than speak to somebody. So we give both options and I would really recommend people to visit the website and either play or ask for a call.
Speaker 1:Fantastic. We're in the busy event and year-end season. What are you looking forward to? Any travel or events or meetings, meetups ahead of you?
Speaker 2:I'm honestly looking forward to the two product launches that are ahead here at Froston. Well, we've launched our new generation AI called Aurora almost a year ago and we will be releasing its new version pretty soon, and we have an even bigger launch coming in Q1. I cannot talk about that yet, but that's something that really fascinates me and I cannot wait to launch it. So, those two big ones.
Speaker 1:We'll have to have you back once the big launch is out there and we can talk about it in detail. Tomas, thanks so much for joining Really impressive and important mission and can't wait to see what's next.
Speaker 2:Thank you. Thank you for having me.
Speaker 1:Thanks so much and thanks for sharing all the insights and thanks everyone for listening and watching. Reach out to rawsonai and check out their offering. Till next time, take care, thank you.