Close
Augustinas Stirbis
“Whatever idea you may have, it’ll most likely turn out to be worthless, it’ll change. Investors put their money into people, not ideas. Therefore, I recommend those striving for success in business to get the right people onto the bus first, and only then decide where you want to go. Don’t press the gas pedal until the idea has been tested, based on data, and actually proven to be needed on the market,” said Augustinas.
This piece of advice is in part based on his personal experience, as only the third product created by CAST AI achieved success. Augustinas, who spent 10 years working for corporations like Barclays and Danske banks, shares a clear answer to the question – what is necessary for prospective unicorn builders to take that fateful step into the tech sector?
“Fifteen years ago, Lithuania’s IT sector was like the Wild West – lacking good practices or frameworks to guide us, we simply muddled through. Century-old companies like Barclays and Danske Bank, as well as their IT centres, showed us how things are done, how to assess risks, plan, prepare for the worst, communicate risks to businesses, and convince them of the importance of change. Over time, however, I started to feel that having too many processes and being 100% risk averse in the short term was limiting progress, which, in turn, led to other, long-term problems. I saw more success by pulling down obsolete processes for financial giants. It was fun, for a while, but I also had this nagging thought that I could achieve more if I wasn’t limited by old, antiquated rules. Then I got an opportunity to sit down in front of an empty page and contribute to its future contents,” Augustinas told us, remembering his early days at CAST AI.
This year, CAST AI grew from a single product company to a platform focused primarily on reducing cloud computing costs, optimising resource use, and boosting operational efficiency, namely – the efficiency of managing Google Cloud, Azure, and AWS cloud resources. The Lithuania-based company’s services are now being used by hundreds of organisations globally – from e-commerce and fast-growing fintech startups to Fortune 500 giants. In late 2023, CAST AI was itself recognised as one of the fastest-growing tech companies in the world.
– As CAST AI’s first employee, tell us more about the company’s beginnings. How did it come to be what it is today?
– Leon Kuperman, the co-founder and CTO of CAST AI, found me on LinkedIn. We talked over the phone, where he introduced me to the challenge and the original idea, and invited me to work together. I fully agreed with him on the problem, but finding the proposed solution a little weak, tried to poke holes in it by asking more questions. He liked me having a backbone and giving honest feedback, and I appreciated his flexibility, deep knowledge of tech, and mature handling of criticism.
Eventually, we founded a company in Lithuania, together. I was its first employee, the only person in the office. The main reason for choosing Lithuania in particular was the Lithuanian roots of the company’s founders. By that time, they had already established several successful startups that, although did not grow into unicorns, became a part of Oracle and Comcast. Their past experience allowed them to realise that, right here in Lithuania, we have strong talents, a good work ethic, and compatible values – an excellent cost-benefit ratio.
– While CAST AI’s first products were technically impressive, they evidently failed to meet market needs. How do you remember this period? Since many startups go through the same process, your lessons could help beginners avoid similar misfortunes.
– The idea for our first product was this. Since the servers of public cloud service providers have different power-to-price ratios, we could provide information on how to get the best result for the lowest price. The interest just wasn’t there, however.
Having assimilated the lesson and developed the second product, we noticed that most data centres are located very close to each other and concentrated in the big cities, e.g., London, Frankfurt, and Mumbai. Our idea was to eliminate the differences between cloud service providers and allow businesses to launch their applications regardless of cloud, server type or location. Although, in engineering terms, the product was fantastic, success depended on whether CAST AI’s two predictions would come true. First – our guess as to which containerized application orchestrator was going to dominate the market in the future. There were more than 10 at the time, and even though Kubernetes was growing in popularity, it wasn’t yet clear if it was going to become the industry standard. We hit the bull’s eye – Kubernetes came out on top and became the dominant application management technology.
The second prophecy was related to the price of sending and receiving data to/from cloud providers – while you can send an unlimited amount of data to Cloud without paying for it, extracting your own data back is quite costly. We hoped that this, likely intentional, agreement between cloud service providers would be terminated, allowing the market to radically cut down on the price of data extraction from cloud service providers (egress). Although the European Data Act, which entered into force in 2024, somewhat restricted cloud service providers’ illegitimate practices, it was too little too late for CAST AI product success. Our second prediction didn’t pan out. The egress costs from Clouds are still in place, significantly reducing the financial benefits that our second product enabled for our clients. The market did not accept the product because it asked for too much change from the clients and too much trust in a young startup. In addition, the price of data extraction failed to go down, and our clients’ savings were therefore quite small.
Figuratively speaking, we created a splendid spaceship for optimising the cost of multi-cloud services, described by one of our clients this way: “Cool, but I just need the water circulation system, not the whole spaceship – could you adapt it for greenhouse irrigation?”. I recall the next 3 months as a period where we consumed a massive amount of coffee and energy drinks – and succeeded. The client was happy because we saved him a lot of money, which convinced us that we had created a good product – and a unique one, for it was based on personal misfortunes. Without all that we’d learned by creating our earlier products, we wouldn’t have been able to build the latest, successful one, which, to my mind, has no serious, direct competitors to this day.
Investors put their money into people, not ideas.
– When did you personally realise that CAST AI was going to be successful?
– To be honest, when we realised that our second product wasn’t going to succeed financially, I got scared. Leon reassured me: “We have a strong team, we believe this issue is solvable, and we’ll look for the solution for as long as needed. It took a single big client, which allowed us to see our product working in a Kubernetes cluster with 100 servers, for us to realise that we were generating value. Six months later, we got another client – with a 500 servers-strong Kubernetes cluster. And today some of our clients manage clusters with 5,000+ servers.
– How do you see the future of CAST AI? What are you headed towards?
– Our cost optimization product has become way more universal and effective, but we must also now deal with copycats – this is the best compliment that a product company can possibly get. Nevertheless, to stay ahead, we must, first of all, keep innovating and retain our uniqueness in the market. Secondly, we can’t remain a single product company – we plan to introduce 3 new products already this year. Our goal is to become a platform solving a wide range of cloud-related problems, not just optimising costs.
– Does this mean that, next year, CAST AI might become a unicorn?
– I believe so!
– What makes CAST AI unique? Share with us some of what goes on backstage – from strategic to tactical decisions. For instance, I was surprised to learn that you’ve never had a separate QA department.
– I’d say our advantage results from our product team’s operational model and values, enabling us to overtake most other market players. For instance, we decided early on that we’dl never have a QA department. Responsibility for the quality and stability of the product lies with the engineers – you build it, you run it. The key was to link up the pleasure and pain points for our engineers, to help them understand that we mustn’t ever shoot ourselves in the foot during moments of weakness – for instance, releasing an update on Friday evening just to feel a sense of accomplishment before going home. It’s much wiser to do it Monday morning with a fresh mind and be there to react to potential risks. Our engineers take turns being on-call, but since both they and I are very passionate about getting our night's sleep, we try to maintain the quality of our products and work principles high.
In other companies engineers often get “pre-digested” tasks: “Do X over there”. We formulate them differently: “Analyse the client’s problem X and propose a solution”. Since we hire only top talent, we hope they’ll tell us what to do, not vice versa.
– Your personal experience could set a good example to all business founders and talent scouts. How does one attract one’s first employees? What works?
– Before I answer, I’d like to share a brief story. During my time at Danske Bank, site head in Lithuania thought we had many qualified IT professionals here and were able to bring up entry level managers, but were suffering from a fatal lack of mid- and high-level management personnel. This limited how successful IT sites in Lithuania could be if no one from Lithuania were at the table where important decisions were being made. The same applies to scaling startups. The Danske Site Head created Switch – an internal leadership training programme, which became an excellent launchpad for numerous young leaders, including me. The programme also had an odd effect, however, with 40% to 60% of participants leaving Danske Bank upon graduation; many of them eventually came back, though. It was an effective shock therapy in terms of developing a variety of skills. Right after completing it, I got a call from my future partner.
I believe that top talent is motivated by challenges and the freedom to work on those challenges uninterrupted. Upon seeing a few wins, many leaders fall into the trap of “syndrome of success” (hybris, arrogance) – I steer clear of them as I’m allergic to the “I know best” attitude. I devote a lot of my energy to avoiding the HIPPO effect making sure that the HIghest Paid Person's Opinions wouldn’t be considered the most important – all engineers are included in the product development processes, and have no fear of voicing criticism. In the end, we reward those who create the most value and impact.
– How does CAST AI form its teams? What talents do you look for? How do you retain them?
– As the company grows, we’re looking for new talents: software engineers, data scientists, marketers, salespeople, customer service staff, etc. After hiring 100 people this year, our team now consists of more than 200. All of the company’s product developers work in Europe (at least one-half – in Lithuania), and most of our product sellers are in the U.S.
As for engineers, we’ve never had even a single junior. Given our niche’s domain complexity, we simply can’t afford to hire engineers with little work experience if you consider cost per change and eventual impact. I see no possibilities for changing this strategy anytime soon. We bring in experienced professionals capable of solving the most difficult engineering problems, and don’t overburden them with unrelated tasks (from excess bureaucracy to useless meetings). Top talents want to solve challenges, and learn from each other. My promise is this – we’ll give you plenty of technical challenges to work on without interruption, and I’ll personally make sure there are no slackers on the team. It works.
My way of identifying top talents is very simple – I ask about the hardest technical challenge they ever faced: the business problem and its solution. After a few technical questions, it becomes clear whether the applicant has had any extraordinary professional challenges and whether he/she solved them or just stood behind others.
– Given your perspective today, what advice do you have for startup or tech company founders – what matters the most at the start? What brings success?
– 90% of startups fail and as many as 60%-70% of the product ideas or improvements even of prosperous ones fall short of planned success. A product created through blood, sweat, and tears, great as it may be, is just an expensive lesson if no one needs it. It’s important to focus on solving real problems faced by real people. If your product cures a big headache for someone, they’ll be willing to pay for it. My advice is – don’t rush decisions based on your ego or a hunch, as there’ll only be so many expensive lessons you can afford. Articulate your hypotheses and potential solutions in writing (slides are evil), look for ways to get data cheaper and faster, perform low-cost experiments, and test hypotheses via unfiltered criticism. For instance, at CAST AI we shelve many early-stage ideas if we can’t measure their potential success or don’t have good answers to expressed criticism. Their authors, of course, feel hurt, their egos are bruised, and their efforts not recognized, but they’d hurt even more if they'd lost their jobs due to the startup going bankrupt . Only clear ideas that can withstand criticism and solve real problems are worth implementing. Needless to say, a touch of luck helps too.
