Funding, AI, and Data-driven Engineering

Vitaly Gordon is an AI & Big Data Scientist with a powerful combination of computer science and applied mathematical expertise. As the CEO and Co-Founder of Faros.ai, he leads a company that integrates all your engineering data sources to give you holistic visibility into the entire software development lifecycle.

What inspired the founding of Faros.ai?

In my journey at Salesforce, I stumbled upon how well a company that is so successful runs functions like sales, marketing, finance, and customer success. Everything was optimized and metrics-driven. But if you look at engineering, it is not at all data-driven. Very little data is being used besides, okay, is this ticket open and closed? That's the level of sophistication that we look for. Most engineers are highly technical, so it's not a skills gap they lack, and it's not like they cannot be analytical. It must be something else. So, I studied this problem, and we realized it was a problem with accessing data. We first decided to solve this problem within Salesforce by building a solution that takes data from all the engineering systems. Think about Jira and GitHub. The CI/CD Instant Management System. Bring them into one place and then really use that data to determine how efficiently the engineering organization runs and where the bottlenecks and things like that are. That's what eventually led me to start Faros.

As VP of engineering, I felt I didn't have good answers compared to my peers in sales and marketing. They have much better answers. And I want to be just as informed about my function as they are informed about theirs. So that is what led to it. It's just the ability to understand. If I run a function, I want to know everything about it and be able to communicate that information. That started my journey of exploring what solutions might help me solve that problem.

How does Faros AI solve that data-driven engineering problem?

There are multiple ways to solve this problem. The most important thing is to understand what the problem is. What do we mean by being more data-driven, and what needs to be improved in being data-driven? The idea of making engineering data-driven is not a novel one. People have been thinking about it for quite some time. However, you cannot name a company that significantly changed this process. There is no self-force of engineering efficiency or work day or service. Now, those companies exist for other functions. And the reason is because it's hard. Every company's tech sector is highly unique, right? It's like a unique fingerprint. So you have to be able to connect all the systems and extract all that data. We use many tools because they are great for our productivity and ability to build things quickly. Still, it isn't good for the skills of managers and leaders to understand what is going on because all of the data is across dozens of different tools.

What we do is we take all the data from all these tools and understand what is the development process. If you think about manufacturing, software development is similar. Think about it: you have a plant, and there is a station there, like the product, the design, the engineering, the quality, the deployment, and the DevOps part. You go through all that, and with manufacturing, every single station in a plant gets measured, and you will look at where the bottlenecks are. Do I have enough quality people? And if things get stuck in quality, adding more engineers to develop more features does not make things faster; it makes things slower. We show our customers some insights, like where the ballots are in their process and how they can solve it.

How does AI enhance visibility and identify critical bottlenecks in the engineering process?

There are a couple of parts where AI is needed in the process. One of them is to arrange data. I'll give you a very simple example. When you use GitHub at a company, many people use their own personal GitHub handles, right? And they bring handles they've used for a previous company, and suddenly, it appears. So how do you know who I don't know, Sudo123 is? It's probably a person, a developer. And let's say they have that GitHub handle. You need to figure out who's doing that work. People have different identities in different systems, so you need to connect with other things. Also, if you now use a system like Jira, which is then attached to pull requests in GitHub, how do you even know this ticket led to this code being written? You need to connect all the dots across all these systems, which are not built to talk to each other.

Using data and machine learning can solve this problem. So that's one part. The other part is also thinking about how you know what to look for now that you have a lot of data that was connected. Or how do you even get to your answers as quickly as possible? All the advances made with technologies like ChatGPT, which we also integrate into our product, are like, why would I go and click and use a bi tool when I can ask where my developers are spending their time? Like what is the biggest bottleneck? If you can ask those questions, it is a natural language that would be a far better interface than the interfaces we're used to today.

How can companies benefit from using the Faros AI?

Engineering teams don’t understand that their business is not to write code. Their business is eventually to affect customer outcomes. So, in many companies we work with, and mostly larger ones, if you look at the process, those companies believe their job is done when the code is written. But that is not true because if the customer is not using the code, then it's useless. It's as if it never existed. It's as if it never makes it to production. So we see that a lot of code is being written, but this code never makes it to production. And now the question is why? That is what you're paying engineers to do eventually. You need that code to go to production. We help companies figure out how, like I said, they can get more code to customers faster. In some cases, we show that there is even a metric called lead time to production. How long does it take an engineer to finish writing code before it goes to production? This efficiency exponentially benefits your ability to compete in the market.

How did you go about raising capital?

So, when we raised our seed round, we had an idea deck. We had a good team of three co-founders who all spent a lot of time in the industry. We were able to raise on that because most seed or pre-seed investors know that whatever your idea is, it will change five times in the company's first year. They should focus on the team and whether it will succeed in making things happen. That is one piece of advice I would give to younger founders who have just started their careers, especially at the pre-seed stage. What matters most is your team. And why do you think that team will win? My example is, hey, I was a VP of engineering at a large company, and I'm working on a product for VPs of engineers in large companies, right? So, it was the founder's market. Fit was pretty clear there. And you tie it with a history of prior success. Those are the things that should be emphasized more in the pitches. Why is your idea great, or why are you the right person to pursue this idea? That is where you probably need to spend most of your time. One of the things that was helpful for our team to raise is, like I said, the history of the team and our connection, the value and the fact that people knew us, and all that.

“Being a founder is a very humbling job. You will have to beg, borrow, and steal every day.”

Why aren’t early-stage founders seeing an increase in available capital nowadays?

Here's the reality. If VCs are sitting on money, VCs are charging a lot of money to manage that money. And they need to spend it like they're not a checking account, right? Like they are given money to spend. If you think about how VCs get funded, they're like pension funds, and those pension funds already have 95% of their money in bonds or public stocks. The VC is trying to get outside returns that could be more correlated with the rest of the market. VCs need to take risks, and that's what these foundations are paying them for. They feel the pressure because if you're sitting on a pile of cash, you must start investing at some point, especially if you are considering raising your next fund. To raise your next fund, you need to show some history of, like, we invested in all these companies, and these companies are successful, and then they got their next investment. So if you wait on the sidelines because again, as I said, it's not that unless there is a significant change in the supply of startups, let's assume for a second there is a fixed supply of thorough ups and there is a fixed supply of money, right? So unless many VCs return their money, the money they collected from their investors, they will have to spend it.

“Founders need to look themselves in the mirror and ask themselves, am I going to get venture-backed?”

Now, the only reason who will blink first is that the VCs will realize they are running out of time and don't have any company they invested in, right? And they need to start investing more. Most VCs don't want to invest in struggling companies. If you were a company about to shut down and were desperate and would take any money, that's different from what VCs are like. They want companies that have very bright futures in front of them. VCs can sit and sit and sit, and there is a period which is probably now where founders find it harder to raise because VCs just keep telling themselves, I have an infinite time; I can sit. Now, we are in this artificial place where it might be harder to raise money, but at some point, the gates will likely open and flood. So, I guess it is temporary for early-stage companies now.

What skills and traits should founders have?

It's a good question. I don't think there is one type of founder, but at the end of the day, every founder needs to understand that this job is very humbling. You will have to beg, borrow, and steal every day. It's a sales job, and you'll probably not be great at it if you don't like it. You sell to investors, you sell to employees, you sell to customers, you sell to PR and journalists. You just constantly need to sell. And most days are pretty bad, and nothing works, and it goes, and then you close the deal, and everyone's happy. I would highly advise people against it if they think it's a way to get rich quickly. It's probably one of the worst ways to get rich.

If these handpicked highlights have sparked your curiosity, you won't want to miss the full conversation with Vitaly Gordon on our YouTube channel or your favorite platform.