James Markarian: The goal with automating integration is to eliminate the latency in the conception of what you’d like for the business to the actual implementation. You can go from the idea of what you’re trying to achieve to the actual execution. That’s what is very attractive to our customers – removing that latency. What we end up seeing with integration projects is everything decays over time. Systems change. Business problems change.
It’s not just the initial implementation of these integrations but it’s the maintenance and the freshness of these integrations which atrophy over time. The ongoing maintenance is actually a huge cost. In the integration space, circa 1990s, tens of billions of dollars were spent on hand-coding these jobs. Through the 1990s to the 2000s, it’s converted to single-digit billions by software vendors that converted a lot of hand-coding into tools that allowed you to build things faster.
Now the model is we should be able to eliminate a lot of the developers all together from that process and do it as automatically as possible. We’ve taken the first step. But then the sky is the limit in terms of how automatic this can become. You’ve talked to Gaurav. You’ve probably heard the analogy that this is the GPS phase of automating integration, but there’s still the autonomous vehicle out there where we could automatically detect the systems that you have and how they ought to be integrated. The whole idea is to improve business agility.
We’re going to start seeing businesses that take advantage of their data being separated from the folks that can’t. This technology is going to be one of the key elements in determining who the winners and the losers are. If you look at data platforms now, it could be Snow Flake or Amazon. They do have some mechanism for onboarding data but it’s the old BI argument. Assume you have the data. Now let me show you what you can do with it.
Through decades of working with customers, we have seen that it is a faulty assumption because that data is actually very hard to come by, very hard to reconcile, and very hard to work with. The goal is to automate that as much as possible so the businesses can fulfill the agility vision that they have for themselves.
Sramana Mitra: When you look around, where do you see open problem for AI to solve?
James Markarian: Within integration, there are two layers at which we work. We work at the meta data level and the data level. At the meta data level, you’re looking at the patterns in integrating various systems. That’s one layer and you can do a pretty good job of predicting those steps.