AI-Powered M&A Life Cycle

The Prime View had the privilege to interview Kamil Msefer, VP of Product Management at Datasite, a company that empowers dealmakers around the world with the AI tools across the entire M&A lifecycle.

Kamil, could you talk a little bit about yourself, your professional path, and how it has led you to your current position?

I am the Vice President of Product Management at Datasite, a leading SaaS-based technology provider in the M&A industry. Datasite was founded about 50 years ago and initially specialized in financial printing. More recently, this has shifted, and now Datasite is exclusively technology-focused. Datasite is working with dealmakers from more than 170 countries, including 74 of the top 100 legal firms and all the top 20 global financial advisory firms. I started at the company as a consultant a few years ago, initially working on some of the new initiatives, and eventually ended up in my current role as Vice President.

Before joining Datasite, I was a co-founder of an innovative real estate lifestyle company that pioneered the concept of retirement communities in Morocco. This company targeted European retirees that were looking for a better quality of life at a lower cost. The project included over 240 homes with many services and amenities. It was a very interesting project as I was exposed to every facet of a very entrepreneurial endeavor. It was not so much about real estate but more about lifestyle concepts.

That experience has served me well at Datasite, as I continued building a team here and making sure that we have a holistic approach to product development. We not only strive to improve for our clients but also to improve the market and Datasite organization. And then, before my latest entrepreneurial project, I was the VP of technology at Intralinks.

Kamil Msefer

VP of Product Management at Datasite

As a VP of Product Management at Datasite, you manage different product teams. Can you describe what kind of people you’re looking to bring to your organization to foster an innovation culture?

Sure. At Datasite, product management is about understanding the key problems of our customers and then prioritizing those problems. One of the main success factors that we look for in terms of bringing in product managers is – number one – people who develop deep understanding and empathy for our customers. We want to go after the problems that have the most meaning and, in particular, will be commercially successful, and then we have to come up with a great solution to those.

We look for people that are passionate about what they do. It is important because that passion ultimately translates into curiosity, and curiosity leads to understanding.

When I say understanding – I mean understanding customer problems. If you understand customer problems in-depth – you are going to figure out how to improve their experience and find solutions. People must care about making the client’s lives better.

The second important thing is that we want people that combine empathy and data analysis. We want people who can process the data, analyze the data, and use it to make better decisions. The final thing – we look for people who embody what we like to call “sense and respond.” You know, a lot of uncertainty is in the very nature of our business. Thus, being agile is key. When looking for solutions to our customer’s problems, the typical approach is breaking problems down into small pieces. We go after the most important ones. We release a piece of code, for example, then we take it out to the market, and we start learning. We solicit feedback, understand how customers are using a particular feature or capability, and then make changes or improvements. It is a very iterative process.

What is interesting about this is that historically before Software as a Service (SaaS – Ed.), software development was kind of like manufacturing, right? You worked on software for a year, and then once you finished it, you package that up onto a disk, send it over to a store, and people buy it. I think what is incredible about SaaS is that we literally can write code in the morning and deploy that code the same afternoon.

We have an amazing opportunity to interact with all of the information at our fingertips. That gives us an understanding of what our customers are doing with our software. It, in turn, allows us to improve software and get better at solving customer problems continuously.

You mentioned the personal qualities of the people you look for, such as passion and empathy. Yet, you also scout for highly qualified professionals who possess AI talent. They are a very scarce resource on the market. Do you offer any data science education at Datasite?

AI is a crucial part of what we do. We have been quite successful at hiring people with the right skills. Some of the key things for us are the basics and knowledge of AI statistical methods, as well as good knowledge of AI libraries. But at the same time, I would say that our approach to AI is practical. Whenever possible, we look to apply existing AI methods to solve business problems. We are not always looking to reinvent the wheel. On the other hand, there are cases when we are dealing with very M&A-specific problems, in which case we will build our own models.

The other interesting thing that I wanted to bring up as we are talking about hiring people is that we have developed a very distributed team. Our core team is in Minneapolis, where our headquarters are, but we also have product people in New York and London. What we have also done, I believe this is pretty unique, is that we have hired people that work remotely from anywhere. That has allowed us to go and find the best talent wherever that talent is. I am an example of this right now since I am based in Montreal, Canada.

Beyond these initial points, we have experimented with a few intriguing programs. We have recently created an internal internship program where we brought people from outside of our AI group and gave them a combination of training and practical experience. Through this, they have been able to take online courses and improve their AI skill sets. We also get them to participate in the actual work. So whether it means helping to train the models, going through the data and understanding the data, doing all that training, or writing some of the code – it all ultimately ends up in these AI models.

Speaking of remote workers, your company seems to have a very forward-thinking model. Yet, do you mean you had remote workers before the Covid-19 crisis, and have you been working remotely yourself?

We have been experimenting with the model of remote workers for quite a long time. COVID-19, coupled with the fact that all of our offices are, at least for now, closed, has effectively translated us into becoming a completely remote organization. Everybody is working from home via chat or video. We have been very successful at it, and we have been able to maintain our productivity.

I see lots of companies struggling with productivity. Maybe you can share a secret on how to keep your teams productive while they are working remotely?

I believe it comes down to a couple of things. One of them, which we use throughout our organization, is the concept of “outcome over output.” Within our different development groups, we are organized as squads. Squads consist of representatives from product management, engineering, UX designers, and delivery managers.

Each squad is autonomous, and at the beginning of every quarter, they set their objectives. They also set their metrics to track progress and decide what features to work on. Obviously, for this to work, the squads have to be communicating very closely amongst themselves, and there has to be a lot of coordination. Fundamentally, outcomes over output mean that each of these squads understands what they need to get done, why they need to get that done, and thus they can work on getting to their objectives. I think that has been a very key element of us being able to work like this efficiently.

The next question is about innovation. Can you explain how you drive innovation in your organization?

For us, innovation is rooted in customer needs and problems. We are customer-focused, and you can hear it in everything I say.

To come up with creative ideas and solve customer problems, it is very important to understand the problems that we are trying to solve and why they’re important to solve.

This belief is not just limited to a group of people, but the whole organization shares it. From the engineers to the UX designers to the product managers, everybody understands the context of why we are working on the things that we do.

From our observations, when you have self-organizing teams with a common understanding of the problems, then you will typically come up with much better solutions due to this collaborative effort.

We have spent a lot of time talking to our customers and looking at the data we collect. Like I said before, we’re very iterative. We will take a problem, break it down, and then in an iterative fashion; we will look to improve and improve, knowing very well that sometimes that first iteration is either going to solve a fraction of the problem or solve nothing. But since we are very iterative and because we bring our solutions to the market very quickly, we can afford to do this. What does this mean for us? It means that the market is also expecting us to be very innovative.

Earlier this year, Datasite surveyed over 2200 M&A professionals worldwide for a report called The New State of M&A.

The research shows that new technologies such as AI are expected to overhaul the M&A process completely.

Another intriguing thing is that due diligence, which is the lengthiest part of the M&A, is expected to be the most impacted by innovation. Our customers expect due diligence to take from 3 months down to 4 weeks by the end of 2025.

This information is crucial for us because it tells us that customers expect us to continue innovating and continuing to drive down the time to get the deal done. AI has a tremendous role in this and is of much interest to us because, when you look at the M&A area, AI can help automate or eliminate repetitive tasks so that our clients can focus on more high-value activities. To give you an example of this – when bankers are setting up deals, a lot of time is spent reviewing and organizing documents. We have AI tools that can look at these documents, organize them, and categorize them into folders. We can reduce the time. It is critical!

Another example of this is redaction. In every M&A deal, you will have to redact content out of documents. The most commonly redacted content is Personally Identifiable Information (PII – Ed.). By using AI and by being able to identify PII information in your documents, and by being able to automate this process, you can save an enormous amount of time.

Now, back to the state of innovation. It is really about helping our dealmakers with tracking documents, people, and status for their ongoing deals. We are looking to help dealmakers get deals done quicker, with less risk and better outcomes.

That’s a very profound answer. You have mentioned the document management system and how it helps to close the deals a lot faster, removes repetitive tasks, and filters sensitive information. So, what are the benefits of specialized solutions like virtual data rooms?

What is interesting about Datasite is that it’s a platform used by dealmakers across the entire deal lifecycle. And as such, it goes beyond simple document management. Its capabilities range from document tracking to organization security. It has workflow and ordinate analytics. Our system is purpose-built for dealmakers.

Datasite accompanies its clients throughout the whole deal lifecycle. When you think about the M&A deal lifecycle, it starts from the marketing phase all the way to post-merger integration. Thus, we built many functions to facilitate workflow or communication between the different participants in the deal. That is why I stress that documents are a big part of it. But there is much more that goes into supporting the M&A life cycle.

The other key aspect is AI. Our AI is trained on specific M&A data sets and expertise that only Datasite has. The ability to make checklists is one example. A checklist from a buyer has a number of things that they are requesting from the seller. The ability to take that checklist, convert it into a folder structure that can be filled with information by the seller is unique to how Datasite approaches this particular market.

Our infrastructure is also important to mention. Our technology infrastructure and our service organization aim to train and organize the best, 24/7, always available for the needs of the M&A industry service and support system.

When you’re servicing the M&A industry, it means that whether somebody calls you at 3:00 p.m. on a Monday or at 3:00 a.m. on a Saturday, you better have somebody out there responding to calls and addressing customer needs. And that’s something very unique to what we do.

Beyond that, due to the way that M&A deals are done, large volumes of data have to be uploaded very quickly. Because of these time constraints, it is typically done in batches. From the technology infrastructure perspective, we had to build our platform in such a way that it could take in large amounts of data in a small amount of time and, on top of that, to upload that content and organize it very quickly.

Another example of a core technology area that is very important to our clients, where we have also invested, is Optical Character Recognition (OCR – Ed.). OCR has to be very fast, very accurate, and has to support multiple languages. And in this, Datasite has purpose-built a platform and an organization that is uniquely able to support the M&A industry.

Datasite’s AI has to process text information written in different languages. Do you implement NLP functionality in-house? Or do you rely on solutions from vendors like Microsoft, IBM Watson, or Google?

As I mentioned before, we are very practical in our approach. We are always going to look to leverage work that has already been done, and then we will look to package it into compelling offers. Here are a few examples. We have a translation feature in our application that uses one of the common AI translation engines. Building our own AI translation engine was never considered. In other areas, like PII redaction, we could have chosen to build our own AI model, but it would make no sense. There are a lot of ready models out there that are very good at finding PII. So, we decided to use one of those and then augmented it.

But there are other situations where we have built our custom solutions. An example that I was going to bring up is document categorization. We talked earlier about our bankers’ needs to be able to quickly go through thousands or tens of thousands of documents they upload. You have to understand what those documents are and put them in the appropriate folder. And so, we built an in-house model that goes through the documents, categorizes these documents, and then gives the bankers or the dealmakers the ability to move them into the appropriate folders. Going back to this “sense and respond” approach, we have built the first version of the model. We took it out to the marketplace. We got feedback. Obviously, the great thing about AI is that you can continuously learn from how your users interact with the documents you have now categorized. Over time, our models are getting better and better and better.

The other part of your question was about languages other than English. One of the interesting things is that when you look at M&A, and this is a global phenomenon, the majority of documents are in English today. But what we’re looking to do overtime is getting more and more documents in languages than English. We are looking to improve and to create additional models that can support languages other than English.

At Datasite, you have different offers for different industries. What are the differences?

Datasite uses the same underlying flexible platform for all industries and services. However, we can customize the offering to suit the needs of each particular industry. For example, we would not necessarily include the same set of features for a health care project that we would for an energy project. We have that flexibility at the platform level to customize a solution.

Beyond that, we have also built a sales and service organization with people who are very knowledgeable in different industries to support these industries through subject-matter expertise. It has translated into us, facilitating over 2500 deals in the health care sector. Our technology is HIPAA compliant, and we are able to process a wide variety of document types, including medical records and other files. Same thing with the investment banking category.

Today we facilitate about 10,000 deals annually, and we are one of the preferred facilitators for the top 25 global investment banks in the world.

The energy and power sector is an interesting one because there is a lot of regulation in that particular area. There are a lot of SMEs that have to be involved in any of these deals. And a lot of these deals are being done on Datasite precisely because we’re able to not only support and process some of the very large and complex files that are in the energy and power sector but also because we can have all the different subject-matter experts access the documents. We’re able to put in permissions. We are sure that people can only do the things for which they have been given explicit permission. So we’re able to support all these different industries using just one platform.

OK, one of the last questions. Do you use modern data science approaches like Deep Learning to build AI models?

With AI, I am going back to the practical nature of how we work. We typically use the best in class tools for everything that we do. For example, when we talk about deep learning, we will use Tensorflow. As you know, Tensorflow was developed by Google to make building neural networks for machine learning easier. Along with Tensorflow, though, there are many libraries built by a huge community to help companies like ours use Deep Learning. That is exactly the approach that we take now for the types of problems we talked about earlier.

For example, with the document classification, those problems are very complex. We are talking about categorization, and we are talking about a multi-label classification across almost a hundred different categories. We have to go through these documents and identify which categories to assign to them. And we’re not talking about one, but multiple categories. And I think the only way you can do that is by using deep learning as the underlying basis. Deep Learning is an integral part of our AI set of tools.

The last question is about COVID-19. What was the impact of the COVID-19 crisis on Datasite? Has the interest in your services increased or reduced?

You know, like for every other organization initially, for us, it was a shock. It was a shock for many people to suddenly find themselves working in a situation where they had to have their personal lives mixed in with their professional lives. But I think this is a testament to the human spirit. What we found is that people have adapted very well to the situation. As I mentioned earlier, we have been very successful at maintaining our productivity levels, continuing to be very innovative. What is also interesting is that we have seen the same thing with our clients – M&A deals have continued. Bankers – who also found themselves in situations where they were working from home – have also adapted and have continued doing deals. And so the whole sector has continued going forward, and we’ve been very fortunate in that we’ve continued operating in the same way. So, despite COVID-19, we remain optimistic.

Kamil, thank you for your optimism and the in-depth conversation about product development in the M&A industry.

Stay tuned for the next interviews!

The maximum upload file size: 100 MB. You can upload: image. Links to YouTube, Facebook, Twitter and other services inserted in the comment text will be automatically embedded.