Andy Sen, CTO of AppDirect, recently talked with ChannelBuzz about some of his views on the industry and where it is going.
Q: How is AI changing things in the enterprise? Are the changes primarily smaller companies or smaller organizations, or large companies doing smaller projects, because that’s what they feel they can manage at this stage in time?
Andy: What I’m talking about is happening in medium and large companies, and it’s bubbling up from the bottom. It’s not an explicit strategy. It’s just happening because people are seeing that they can build tools for themselves. And from a IT perspective, you really need to manage this process. So you want to encourage people to use AI to build tools. You want to give them some kind of a playground where they can be successful. You want to give them access to AI tools and LLMs or anything else which can be controlled or monitored so that you’re making sure that that company information isn’t going out to the cloud. But I think the actual development from the ideas, the execution is happening from the bottom up.
I know that’s not intuitive. Ordinarily, you’re talking about smaller projects. And so a lot of people are going to come to a conclusion that it’s smaller companies. It’s happening at all scales in all different companies, and it’s happening because in many cases, just somebody sees work that they’re doing or some system that they’re running, and they might have some kind of top-down mandate to save costs. And they’re seeing that they can build quick point solutions using AI. Now, you know that 25, 30 years ago, we had a similar phenomenon when Microsoft came out with things like Visual Basic, where small tools were being set up very quickly. With AI, a similar kind of projects are being done, but at a much faster rate and by people who frankly have, would never even think that they could build a tool for themselves using an AI program. They may never have programmed before in their lives, but using AI, they’re quickly able to create tools which can help.
Q: How have IT departments responded to these developments?
Andy: I think this has, frankly, caught a lot of IT departments by surprise, and now they’re trying to figure out what are some ways where we can govern these things. Because the business impact of these small tools is great. It’s saving people money, it’s giving creating more intuitive products. But from an IT department, the question is, well, how do we support what’s happening on the ground? This isn’t programmers, per se, who are doing this. It’s line of business people who are able to do this because of the combination of having the tools, and the improvement’s AI.
It’s the consumerization phenomenon, and it’s probably starting with people using things like ChatGPT and their personal lives. They may then, uh, hear from a friend of them who tells them about Michael, and that someone’s downloaded something like a a cursor, spoken a few words and come up with this for the software. All these examples that I gave you, even within our own company, were implemented not by software engineers, but by lines of business. And now there’s a second thing that as these tools start being created, there is an opportunity once again to take these tools and monetize them. Maybe somebody in your line of business has done something, which would be valuable in the external world and can be monetized. Now, I haven’t seen examples of that immediately, but that seems to be what would happen next. Again, all of this requires some kind of an environment where the IT or the executives of the company have visibility of what’s happening, and then they can put their thumb on the scale. They can encourage some of these tools, maybe discourage some things which are using up too many resources, and, you know, even promoting from monetization, some something which is just very successful.
Q: Isn’t there concern about the shadow IT phenomena with this that people will, essentially, go off and do their own projects without necessarily going through the normal chain of command?
Andy: Absolutely. I mean, this is shadow IT on steroids, and that’s why you need to balance it. The reason you have shadow IT is because pretty much every organization, the list of things that you want IT to do is enormous, and that’s never being done. With the advent of AI, more and more of these tasks or wish lists or roadmap items can be done by the lines of businesses themselves. They will end up doing it. That’s why I keep coming back to this idea of, instead of IT, looking at this giant list of to-do’s or wish list, roadmap of years, and saying, okay, we’re going to take this one by one, maybe accelerate within using AI, until what we thought we could do in three years would be going to get done in one. I don’t see that happening. But what I do see happening is IT providing some kind of an environment where it could be buying like a corporate subscription to something like an open AI or an Entropic. Some way of really enabling all this IT happening in non-IT departments, but having the visibility over it, having certain guardrails over it so that it just doesn’t go crazy, I mean, someone doesn’t buy $10,000 worth of tokens or create something using an unauthorized LLM where the company data is pushed out. So I think needs to have that visibility around it, but they’re not the ones actually building the tools and applications. I mean, it’s all the trends of shadow IT that’s just become accelerated. And now what IT needs to really do, is to make sure how they can harness the situation in this new world that we’re in. You’re seeing this and changing at some point. where it does become more of a controlled phenomena, because the upper level people want to keep their thumbs on the scales.
Q: Do you think these changes bring a democratizing effect with them?
Andy: Absolutely. I think it needs to be more controlled. If you don’t have any control, you can get into all kinds of financial and compliance issues. But I think this idea that the work that traditionally was done by IT is now becoming democratized or consumerized and spreading to other lines of businesses. I don’t see that trend going down. It will happen. It is happening now, and it’s up to IT to really control or to have visibility over these efforts. But I think they won’t, it’s not a question of them saying that, ‘don’t worry about the effort. We’ll take on the effort ourselves.’ I don’t see that happening, but I definitely see the need for them to have control and visibility. If you look at any of the commercial ERPs, they’re pretty amazing pieces of work. The amount of functionality that they can support is enormous, because they’re trying to be everything for everyone, regardless of what industry you’re in, regardless of where in the world that you’re operating, they have to support all those use cases. And supporting all those use cases can make them clunky, whether it’s from a user experience point of view or setting up the workplace point of view, it’s a lot. And every single company maybe uses 20%, 30% of all those capabilities. So, either the ERPs will have to adapt from a licensing perspective and have very specific costs and license models to allow people to pay for the 20% of the usage that they’re using. Or they’re going to find more and more competition from these small tools that people are building. People are building small tools, and then they’re even more reluctant to pay the traditional licensing costs of these ERPs. So I think there’s going to be a lot of competition in that market. And it’s competition for the ERPs, but it’s definitely an exciting time for them and for everyone else. It gives opportunities for people to have really focused point solutions. It gives an opportunity for new licensing and pricing models to come up. It gives an opportunity for CIOs to really be able to align what they’re paying to results.
Q: You said organizations still won’t be able to measure AI’s impact on the bottom line. Why not?
Andy: The context for that is that it takes time for AI to fully transform a business, until you can see it in like the bottom line metrics of revenue or cost or things like that. The first effect that’s happening is it’s increasing efficiency within the company, and increasing the scope of what an individual department can get done. You can measure those things, and you absolutely should measure it, both the departments as well as the IT should be measuring it. But it’s probably not going to show up in the balance sheets of companies, at least not for next year. At some point, as this transformation gets bigger and bigger, it’ll become obvious. So the analogy I like to use is that of the Internet. When companies first started using the internet 30 years ago, they even threw up with their 1st website. Maybe they just began to figure out how to accept orders online. Those particulars, I mean, they got more impressions, they might have a little bit of online sales, but it probably didn’t affect the sole bottom line of the companies until a couple of years when the internet was used everywhere. And I think artificial intelligence is following a similar trend except it’s happening much faster. Next one seems pretty self evident. The future of AI innovation is talent fluid. You no longer need to be an engineer to innovate. If you look back even five years ago, there were very distinct roles, there were project managers, there were designers, there were engineers, and you needed a team of such to build anything usable. I’m not even talking about building a massive piece of software, just building a smaller point solution. And more and more of that we see being compressed by talented people, who do the entire thing from design, to testing, to deployment. All of these steps can be helped or even done to like many percent by AI.
Q: Agentic AI is complicated. It’s something that people are doing, but a lot of them don’t really seem to understand it. Have you built that into your projection, that these concepts which are being used, and the tools which are being used, are simply a lot more complex than they were two years ago?
Andy: Again, you start with things which are simple. We start with just the LLMs to make something like a chatbot. Now, anything like things like cursor or anthropic, they have agentic models to help you build software. There are many companies, including the company I work for, just trying to create frameworks for making agentic development easier. And it’s one of those things where the tools get better and better. No single tool right now is probably going to take over the world and that’s the one standard for agentic development. There’s a lot of tools out there, which are good, and it’s all a matter of threshold. At this point, I’m just throwing out a number, let’s say, in a company, only 10% of the employees are curious enough or interested enough to use these tools to build something useful around authentic AIs. Next year, it will probably be 15%, 20%. So as that, and you never have to get to 100%. If you can get to, I don’t know, 20, 30, 40% of the uh, of a company being empowered to create their own solution using any of these uh, agentic tools or any of these tools that are provided by vendors. You’ve got, I mean, you’ve got a revolution, you’ve got a completely different type of company at that point.
Q: So in terms of what is going to happen in terms of AI – are we going to see a revolution? Or are we just going to see more incremental growth?
Andy: With the first, and for that matter, second waves of AI, where you had growth, it wasn’t like it is now. I think what is happening now is fundamentally different from the first couple of ways. I think this time, it’s different. And I think the reason I believe that is, I’m seeing people not from a traditional IT background or software background actually using AI, to do useful work. And this wasn’t true before. It’s not just, you know, creating like funny pictures or clips. So what I see in the next year, It’s both. It is incremental. It’s incrementally more and more people will start using these tools. And it is revolutionary because that’s a revolutionary idea. The idea that an increasing percentage of your employees can now build their own tools. If you think about it that way, and I don’t know what the threshold is to say that this is a revolution versus increment, because we’ve just never been there before. So I’m personally very excited about the trend. I think it’s real and I think it will be revolutionary.
Q: What are the channel implications of all this?
Andy: From a channel perspective, the channel is trying to help their end customers, or advise them. I would say that just making them aware, of all the tools that are coming in, I think that the choice they will have in the channel for different tools and just different products that they can push through will increase enormously. I encourage everyone to just know about, know what’s coming, and an important skill will be how to combine these to create a compelling solution to whoever you’re advising or selling to.