Episode Transcript
[00:00:03] Speaker A: Welcome to INA Insights, where prominent leaders and influencers shaping the industrial and industrial technology sector discuss topics that are critical for executives, boards and investors. InA Insights is brought to you by Ina AI, a firm focused on working with industrial companies to make them unrivaled. Segment of one leaders to learn more about Ina Aihdev, please visit our website at www. Dot Ina dot AI.
[00:00:40] Speaker B: Hello and welcome to another episode of INa Insights podcast. Today our guest is anti Karyala Nen, who is the co founder of Sema Four. Sema Four is redefining knowledge work by enabling enterprises to build intelligent agents that enhance decision making and productivity. The company has raised around $55 million in funding to date. Before Semaphore, Andy was the founder and CEO of RoboCorp, which later merged with Semaphore. He has a strong background in open source innovation and now focuses on advancing automation as code, AI actions and generative AI at Semaphore Auntie, a very warm welcome to our podcast and we are very excited to have you here today with us and are looking forward to talking about your journey as a three time founder.
[00:01:34] Speaker C: Awesome. Thank you for having me.
[00:01:36] Speaker B: Perfect. So let's start with talking about semaphore. Can you tell us a little bit about what the company does and what problem is it trying to solve in the AI space?
[00:01:47] Speaker C: From the beginning, we have been focused on AI agents. Now, this company started earlier this year, and what we are basically doing here is leveraging all the fundamental new capabilities that generative AI has brought to us to enable new types of enterprise business process automations.
So essentially building a platform where companies can build, run, manage enterprise AI agents and also empower business users to authorization and control the agents.
[00:02:21] Speaker B: Got it. And how do these AI agents help the different enterprises? Could you give us some examples?
[00:02:28] Speaker C: So agents are fairly flexible and they have incredibly wide ranges of use cases. So initial use cases are around things like customer support, internal operations, driving efficiency. So basically taking workloads that people have been doing manually routine work and turning that over to AI. It can be things that require you to move between multiple applications, like in customer support and service work where you have a customer on the call, and you might have two, three, four different systems where you need to pull in information and then modify information in. So the agents can help you either pull in the information and then also take actions actions on that information. And so that's really the early set of use cases, kind of taking where we started with robotic process automation, RPA and various other automation tools, and then enabling AI to broaden the scope of that, and then next stages of that would be around compliance areas. So for instance, in finance, we have a lot of workflows that are fairly complex knowledge work tasks that we can delegate over to AI in the compliance space. And then I think that ultimately we kind of go into inventing new streams of revenue enabled by AI agents. So going from saving to protecting cost and a reputation to growing revenue.
[00:03:55] Speaker B: And are there any particular industries where you're seeing higher adoption versus some of the others for these AI agents?
[00:04:05] Speaker C: AI agents are extremely horizontal technology, so ranging pretty much from industry to industry. We are focused on functions like the CFO's office, HR, it teams, support teams, etcetera. So it's basically functional instead of industry focused. But now in terms of adoption, there's incredible demand from banking, financial insurance. But again, those are so highly regulated that the pace of movement is slightly slower than with for instance, industrial manufacturing, oil and gas.
Those tend to move faster. So we see interest across the board in terms of industries, but certain industries are moving just faster to adapt first use cases than others do the regulations.
[00:05:00] Speaker B: Interesting that you mentioned in industrial space, aunty, because that's one of the industries that I now focuses on. And it's great to hear that the adoption there is faster, because typically the common knowledge is that industrials or the manufacturing space is slower compared to some of the other industries. In terms of new technology adoption, what trends are you seeing in that industry with respect to adoption? I mean, you mentioned a bunch of functions where the AI agents are applicable, but where is it that you're seeing most opportunity and adoption?
[00:05:33] Speaker C: Yeah, so we have been tackling initially a lot of use cases that require processing complex unstructured documents and typically reconciling that data against back end systems. So we would see customer order processing and accounts payable, accounts receivable, these kind of flows that might require email transactions between customers or might require PDF documents. Like we've seen use cases where we have 100 page PDF's that need to be reconciled against accounts receivable systems, for instance. And we can take those PDF documents that used to take a human person multiple days sometimes to reconcile, go through line by line to match against an ERP system to see if the if line items match, and take that down to matter of minutes to actually automatically reconcile with an agent. And the agent basically takes in the documents and says that, okay, I found four lines that have discrepancies. Can you help me reconcile them? That's a generic level description. I know, but when we talk about industrial companies, for instance, I think a lot of that customer interface work, whether that's getting paid or paying invoices to vendors and, and that sort of thing, there's a lot of opportunities that internal teams are just by the habit now doing manually, and it's something that you can, if your business grows, you have to grow your internal teams by hiring more people. Now we have a new technology to agent that we can actually delegate some of that work and we can grow smarter by not having to add headcount.
[00:07:16] Speaker B: Interesting. And as semi four is doing this, it has an open source approach. Right. So how is that contributing to, let's say, democratization of AI and some of the use cases or applications that you are building?
[00:07:32] Speaker C: Yeah, so open source has been fundamentally important to me personally as a founder. I built my previous company around open source and semaphore also, all of our founders have that deep background in open source technologies to companies like Cloudera and Docker. And so open source in AI currently is one of the cornerstones. I think the space is moving so fast and the speed of innovation is truly tremendous.
And building on an open source foundation allows us to tap into that breathtaking speed of innovation in AI technologies. And I think there's certainly interesting things happening on multiple levels. Like let's start from the foundational model level. Meta has released just the latest set of AI models, Lama 3.1 as an open source with an open source license. That really firstly puts a lot of pressure on the closed models in terms of competition. That allows companies like us to take those open source models and fine tune and adapt them to use cases like ours and make them that much more efficient. So it's really enabling a lot of innovation. On the developer framework side, we are leveraging a lot of the top open source projects in AI frameworks to enable the speed of development that we have. And then on top of that, we are building our own frameworks as open source. So it's really a multilayered approach where we leverage innovation from the foundation model upwards. And that's ultimately what we need to do as an industry to make sure that this innovation gets to the hands of customers faster and faster.
[00:09:17] Speaker B: Got it. And are you seeing a greater and greater trend towards open source across different players, across the entire value chain of the AI industry?
[00:09:31] Speaker C: Yeah, I think so. A lot of the innovation is being driven by open source technologies. Now, as a business end user, do you really care that it's powered by a stack of open source software? Maybe, or maybe not. I think fundamentally what it brings you. Is this sort of a peace of mind not to be tied into, let's say, a model vendor? You have the freedom of flexibility to change model vendors.
When we have an open ecosystem powering the technology, you're not tied to, let's say OpenAI on the models or Azure or Google. You have a choice of options there.
[00:10:11] Speaker B: Right. And I suppose that this open source nature is of course leading to more collaboration within the different players. How is it impacting the different partnerships that could be formed within the AI industry?
[00:10:26] Speaker C: Yeah, well, I mean the space is still moving very rapidly and it's on companies like us to make sense of all that rapid development. So partnerships are formed, dissolved new partnerships created, so it's all in very ravaged flux. But at the end, end of the day, we really don't want to expose that to the end customer. What we want to make sure is that our customers, who are line of business experts, they have a great experience to describe the work that they want to have the agents do for them and get that work translated into actual working, functioning agents. And beneath the surface there's all this crazy new models coming out and new capabilities being unlocked by the month and by the day.
[00:11:14] Speaker B: Right, right. That's very true. We hear about them every day. Auntie, I would now like to focus on your personal journey. Could you tell us a bit more about your background and your life before semaphore?
[00:11:27] Speaker C: Yeah, good question. Very interesting one to switch to.
Like, my journey started as a software engineer. Like many, many tech entrepreneurs, I think, you know, I sort of wandered into entrepreneurship, to be honest. I never was set out to as a passion thing to do that, but it sort of was easy to gravitate towards. Like, I think it's one of these last adventures that you can take in the, in the modern world to start something new and build something from nothing, essentially. So it was sort of just a form for me to gravitate and work on things that I'm passionate about. So ultimately that led into finding out how startups work and how venture capital works and just the passion to build new things from nothing, basically.
[00:12:20] Speaker B: And what inspired you about the idea for RoboCorp, which is now semaphore, of course?
[00:12:27] Speaker C: Yeah. So my background was in Python and open source projects around Python, and there was this explosion in robotic process automation, say around 2006. A lot of the leading companies got really started.
They had been started earlier, but they really took into a lot of growth around that time. And I happened to bump into RPA as a category around that time and I knew that this some cool open source technology that we could apply to that problem and solve it in a more elegant, better way.
And it felt like I had the combination of understanding and actually finding out about RPA and then having the knowledge of these technologies and these communities in open source that I could bring together kind of naturally. And I remember the day very clearly that I thought that, well, if I'm not going to build this, nobody might build it. So guess I had to just do it. And so I started kind of gravitating towards that problem area and wanted to build a project around it, which eventually became a company.
[00:13:37] Speaker B: And you've been a founder three times now. What have been some of the challenges that you faced during this journey?
[00:13:45] Speaker C: First time founding a company is sort of eye opening in a way and it's really exciting, but that company was just sort of a boring old consulting services company. But it was still like, it's fun to build something from nothing. Start growing it, hiring people, selling your offerings, which was service at that time, and getting new customers, hiring more people, getting new customers, hiring more people, and sort of growing that and I, and understanding the dynamics of it all.
That business got acquired after, I think it was three years. There was a conscious decision of whether we should hire management, some middle management to scale it out or then sell it off. And we decided, well, I guess we've seen this movie already, but there's nothing too exciting to build beyond that from a consulting company. And then Robocop became a proper product project and a company and VC funded. So that was altogether a different journey. Of course, it's very different to build a product and take it to market and figure out enterprise sales and marketing and customer success and growth and scale, rather than building a services company. So Robocop really has been the company that where I learned a lot of the stuff that I know today. And now starting semaphore, it's really great to work with co founders who have all been through launching companies, building new products, building go to market organizations. We have extremely talented team behind the company and it's sort of getting to do this with the knowledge that you have from your previous experience and mistakes as well. And getting it to do sort of proper is a great feeling. Although like, I think we did a lot of things intuitively right at Robocop, but there's always that learning curve that you need to go through. And now being able to apply that learning again is a great feeling. And doing it with a team that's extremely talented, I get to learn from them as well. So I think it's, you know, like I said, one of these last adventures that you can have in the modern.
[00:16:10] Speaker B: World, I think, yeah, excellent and anti. Any advice for budding entrepreneurs from your experiences and learnings?
[00:16:19] Speaker C: Yeah, so one of my hobbies is coaching and helping other entrepreneurs. Whether I'm an angel investor or not, I typically try to give sober, good advice to people, and somebody might be thinking about, hey, I have this idea. I've been thinking about starting a company for a few years. Like, should I do it? How could I do it? I mean, it's never easy to take the jump, I guess, to launch something, and you need to tolerate a lot of risk to do that. But I've done it a few times I've never regretted.
And I think whatever happens in the journey, you learn so much compared to the previous thing you were doing, that it's typically, to me, has been always worth it. And I think that applies to a lot of people. And then during the journey, you know, I know a lot of entrepreneurs who have grown their companies to billions in value and hundreds of millions in revenue. And it's fun to meet with these people.
When you have lunch with them and they're like, oh, my God, again, I have this headache. A problem. You realize that it's in different times during a company's journey, you always have a set of problems and challenges, just shifts the area where you need to focus on where the challenges are. But it's fun to see, like a CEO of a multi billion dollar company having sort of the same type of struggles than somebody who is just launching their company. It's also a never ending thing, but as an entrepreneur, kind of want to gravitate towards those kind of situations. I think it would be really boring at some point if you don't have a problem to solve in your hands, always.
[00:18:09] Speaker B: So, Andy, just last question to close this. What's next for SEMA four? Especially if you could put some of this in context of all of the buzz that we are hearing around generative AI.
[00:18:22] Speaker C: Yeah. So I think this year will be the formative year for AI agents. So we, we saw the genetic AI expulsion last year. Started from the chat GPT moment and people figuring out rag and these knowledge assistants and copilots.
And now people have been exposed enough to the technology that we truly understand how we can put it to use in many very practical and big problems. And AI agents is one conduit to that. I think this year will be the year of agents, and I'm looking forward to being able to, soon enough, talk publicly about some of our first enterprise customers going into production with our technology and the use cases and the value that they're seeing. So that's really something that I'm looking forward to. And obviously being able to kind of launch products publicly, and there's a bunch of companies racing towards that public launch and around agents. I think this space will be incredibly big. It will be one of these, you know, meaningful, big shifts in enterprise technology and, well, technology more broadly. But I'm focused on enterprise productivity. So to us, this will be a major, huge new unlock for a lot of value to be created down the road.
[00:19:42] Speaker B: That's very exciting, aunty. And good luck for the public launch. All right, thank you so much for your time today. Anti. We really appreciate this and it was a great conversation. Thank you.
[00:19:53] Speaker C: All right. Thanks for having me.
[00:20:01] Speaker A: Thanks for listening to Ina Insights. Please visit Ina AI for more podcasts, publications and events on developments shaping the industrial and industrial technology sector.
