Interview with George Roth: Innovation as critical mass, the AI revolution and looking beyond tech capabilities

Meet George Roth, a Romanian emigrant whose story and journey are worthy of a cinematic portrayal of ‘ The American Dream’. As AI Evangelist at UiPath, he has significantly shaped the tech-powered business landscape through his innovative ventures and his pioneering dedication to tech-mentorship.

Explore this compelling read and empowering interview with George Roth, known for his visionary approach and strong commitment to addressing the creation of unicorn companies while staying at the forefront of AI innovation, spanning topics from entrepreneurship to sustainability.

What is it that keeps you motivated and driven in such a fast-paced industry like technology?

I’ve been in the tech industry my entire life, drawn to its dynamic and ever-evolving nature. It’s a field that is almost always exciting and rarely boring, though it’s important to look beyond just the technology to the showing of its broader impacts.

But at the same time looking strictly at tech, you lose some very big things in terms of the context of what’s going on around tech and what tech influences around us.

This generation is living and morphing with tech, which in my opinion counts as experiencing a new industrial revolution, driven by the widespread adoption of generative AI.

This will make the tech industry perpetually advance and also bring new challenges not only for the people working in tech but also for humanity in general. So I am super excited about being involved in tech continuously. I also have a job that allows me to read a lot and to contact a lot of companies who are doing pioneering work in the generative AI space, so tech is exciting, but always be with one eye around tech and make sure its implications don’t hurt what’s going on around it.

One of your qualities is finding strategic partners for your UiPath. What do you look for in a potential partner, and why?

That’s one of the things which in my entrepreneurial life I neglected – the partnership path. And unfortunately, many of the entrepreneurs have the same tendency to try to do everything by themselves, which is impossible. So I learned I had a chance to work in a Technology Partnerships group. And I learned a lot. First of all, there are multiple types of partnerships from a startup point of view.

For example, I met a company that was working on a system to store and distribute medical data, and medical information of the patients. It was a fantastic idea, and the company still exists. But at a certain point, they told me that they started to do a system to do automatic diagnosis of lung cancer. They said, ‘We know how to distribute images of tests and thought we’d add this functionality.’ I told them, ‘You’re forgetting your main goal. Your vision is to make patient data shareable among doctors. And other companies are specialized in doing this. Why don’t you partner with a company that does this?’

So first of all, partnerships are super important to maintain a startup on the main course of what they are working on and not to derail it, and this is where technology partnerships come in. Now the second part, the business partnerships, are super important in go-to-market and can be classified into multiple categories.

One category is the partners who help you to resell, which is important because building a sales team is one of the most difficult things a startup can do. First of all, you need to find trustworthy salespeople, which in itself is a big challenge when you’re not willing to invest tons of money and wait for six months or a year to see results. That’s the first challenge, and finding a reseller partner in what you are doing helps you a lot.

Another type of business partner helps you implement your product for different clients, which is crucial for B2B companies. Many startups, including several in Romania, have founders who also handle implementations. This is risky as it can lead to deviating from your roadmap to accommodate a partner’s requests. Partnerships with professional services firms are ideal—they can handle implementations and often combine reselling with professional services, making them the perfect combo.

But there’s a fine line between pivoting & taking advantage of opportunities and staying on track with your roadmap. How do you walk that line?

Pivoting is one of the most challenging aspects of business, especially in fast-evolving industries. At UiPath, where I work, I’ve witnessed pivoting on a large scale. It’s amazing to see how it evolves and how it adapts to how the market changes. This being said, one must carefully balance seizing new opportunities while staying true to the strategic roadmap.
This involves thorough evaluation and cautious decision-making to ensure sustainability and relevance.

For example, the Robotic Process Automation (RPA) market, which is very close to me, was the new hot thing around 2018. However, it gradually became a commodity. Companies that relied solely on RPA are struggling because there isn’t a standalone market for it anymore. Similarly, in the Intelligent Document Processing (IDP) space, which I am also familiar with, simply having a product was once enough to be successful. Yet, as these technologies become commoditized, you must pivot towards offering complete automation solutions that solve real problems end-to-end. Pivoting is essential throughout the lifecycle of a startup, especially in fields with significant existing competition. You must adapt to stay relevant.

Speaking of commodifying technologies, which specific emerging tech trends do you believe will have the most significant impact in the next few years?

One of the big trends is the democratization of tech. Tech will become more and more open to people with lower technical skill levels, but it will increase the requirement for people with multidisciplinary skills. So for example, tech will be available to biotechnology engineers who were not trained specifically on AI because the level of skills required to use the technology is being reduced all the time. With generative AI, programmers as we know them will disappear. You won’t have to know how to write code in a certain language, but you will have to know how to formulate the problem in a way that the computer will understand it. There’s this new trend called LAM – the Large Agentic Models, which facilitate problem specification in everyday language and generate workflows automatically.

The second big trend that will happen in technology is the implementation of constitutional AI, this term introduced by Anthropic means you provide a system to somebody to be used, along with your constitution of what not to do with it, an idea that already exists in contracts related to AI tech.

The third one is that generative AI will become a commodity and, as a colleague of mine says, we’ll experience invisible AI, as people won’t know when they’re using AI.

Another big trend, staying in the AI space because that’s my expertise, is moving to the edge, as we’ll see more systems operating independently of the cloud, crucial for applications like IoT
where immediate, local response is necessary. The edge means that you are working with systems that are not connected to the cloud to reduce costs and the latency of responses. In robotic surgery, for example, you can’t wait to see a latency, and these devices will locally store the knowledge. Therefore, the LLMs (large language models) and the LAMs will be reduced in size to be run on the edge.

Of course, blockchain is a big trend that will be used more to ensure the integrity of the data and to democratize different services; definitely, the new quantum computers will have to increase the speed of processing. And then if you take technology as a whole, the energy sector will develop, and so will medicine. Open source is, without doubt, important to the whole process, without it it will be impossible to make any long-term progress. After this new industrial revolution, the effects will be seen across society.

Are we going to be ready to manage or regulate these types of very advanced technologies as they advance?

The danger of regulating tech is that politicians are getting into this space because it’s good for populist politicians too. It’s a subject that if you look in the U.S. there are two trends there is one trend people say oh the AI will come and will kill us, and the other view says we know when we develop the AI we have guardrails too and the AI is so smart that we can prevent to do bad stuff.

I recommend everybody who has a chance to look at the testimony of the U.S. Congress of three scientists, where one of them said through the regulation we should force every cloud system that runs AI to have a button to be able to unplug the whole cloud if something goes wrong.

But when it comes to more realistic discussions, probably Europe is the most advanced in its regulation framework. The danger is that we’ll introduce some measures that will make it more difficult for companies to develop AI. So the fine line here is not to over-regulate the scene. In the U.S. it’s a little bit different, as there is no centralized AI. AI legislation is practically the legislation for each vertical slash industry, in every state.

This process of regulating AI will go recursively, with restrictive policies that will harm the startup movement, as it will require a lot of money. Because of the prohibitive cost of big AI, which means that to train models you need computers that cost millions and millions of dollars to train, you can risk creating an elite group of large companies that control the industry.

And we want to avoid this monopoly of too-big-to-fail companies. When you own a startup, you have to think in terms of your dependency on big tech. If I’m developing an LLM, that does something fantastic, never done before, I won’t afford to train a big LLM, and I’ll do the reduced version of it exclusively for what I need.

There are challenges for the startup world, first of all, to stay in a certain niche. My advice here is don’t try to solve huge problems because ultimately, if you succeed, you’ll end up depending on big tech giving you a billion dollars from one pocket and you’ll pay them back when you train your model in the other pocket. Ultimately, big tech will control a lot of the process and I think what has to be done is to regulate the big companies, and not the small ones.

So it’s an exciting time for technology in general. What advice would you give someone who’s just starting up with a tech startup?

There is a recent classification of startups by Sequoia, called The Arc Product-Market Fit Framework and it classifies the startups in three categories. You have the ‘hair on fire’, the ‘hard fact’, and ‘future vision’. First, you have to establish what it is you’re trying to do. If you want to go after one of Google’s products, that’s very hard. Because even if yours is better, the cost of marketing to replace somebody is prohibitive. Don’t go after the big companies at the beginning.

If you come out with something disruptive, and future visions start, you have to make sure there’s a need for it in the market. If your product is incredibly exciting but there’s no market need for it, it won’t succeed.

If you do something disruptive, and build it gradually, do it in a way to release as soon as you can to the market, and understand that the MVP is to get feedback.

Especially if you’re working with AI or expensive technologies, make sure you calculate how much the tech and the staff costs, then take a look at the market and very clearly define your category. And of course, the third piece of advice is to get advisors who went through this process before and who know what you shouldn’t do.

What steps should be taken for the Romanian startup ecosystem to stay competitive or even to become more competitive?

Romania is developing very nicely in the sense that there are more and more investments in the seed market. And because there are more and more successful people who decided to get here and to finance which is great, there are a few very big successes, which are used as role models for the companies, and there is also an ecosystem of accelerators and incubators, which is developing.

What Romania could do is increase the efforts to train the people in product and go to market. I think Romania is lacking in product engineers. Product professionals are a type of their own, and an engineer cannot be a product person, by himself, because they are too involved in the technological side of it.

An ideal company has three different sides on the product side: product specialists, people who specify what to do, engineers who implement what the product professionals decide should be implemented, and R&D, which is ideally looking for the future after the product is released.

So each time when you mix these three and engineers start driving the product it’s very dangerous and, unfortunately, Romania doesn’t have enough product professionals. And that should be the objective of every organization that wants to boost the Romanian startup world – to focus on training product people.

The second thing to do is improve go-to-market knowledge, which is also lacking in Romania. Usually, Romanian companies start there, and then they have to learn how to go to another country, often through a slingshot, which means taking what you have in a country and slingshotting it to another country. Don’t try to go to market from Romania globally, because it’s virtually impossible to create a product in Romania and launch it in the world, as you have to find a few launch pad countries like the United Kingdom, the Netherlands, or even the United States.

Speaking of contributing to this new generation, that still in the formation years, you are a mentor of Endeavor Romania. What’s the impact of Endeavor Romania in the local ecosystem and why did you join it as a mentor?

I think Endeavor is a fantastic idea that’s not even new. Everywhere in the world, especially now when people are moving from one country to another, people always have ad-hoc startups coming from a country looking for co-nationals for these solidarity-based interactions.

But what Endeavor did was institutionalize this tendency and to create a framework to encourage this type of networking. Endeavor is huge because of the critical mass of successful people who are willing to give back to the movement, which is such a powerful value that it’s the reason I joined this organization. I was working with companies who are in Endeavor, non-formally before, but Endeavor is a fantastic way to create these global frameworks for startups.