Billionaire Tom Siebel Explains His Fascination With AI

Tom Siebel has been riding the tech wave since getting his Masters in Computer Science in the mid-80s, during the mainframe era, and is still (way) out in front.

He joined Oracle when it was a startup as employee No. 20, worked his way up to SVP, then left to start Siebel Systems, a CRM software enterprise venture, with co-founder Patricia House. In 2006, Siebel sold the company back to his former employer, Larry Ellison, for a cool $5.8 billion, and in a rare move, Ellison kept the name.

But in 2011, when his contemporaries were considering improving their golf game, Siebel stayed in the race, immersing himself in the worlds of natural language processing, deep learning, artificial intelligence, and the Internet of Things. Alongside Pat House, Siebel launched C3 IoT, a software platform company that's disrupting a number of industries, from the military to energy.

As an estimated 50 billion machine-addressable smart sensors get installed by 2023, everything will become a computing device, and Siebel is ready to analyze, improve, and predict efficiencies. PCMag sat down with him before his keynote at the invite-only Montgomery Summit in Santa Monica to find out more.

Let's start with some historical perspective. You've seen the tech industry go through massive changes, including those you initiated. When I first started computing, it was on Control Data CDC Cyber machines, very large computers, and we programmed on punch cards. Then it was the era of mini computers, and I used PDP-8, where you programmed on paper tape. So I've seen some changes. When I studied computer science I became very interested in developing expertise in relational databases. That was my graduate work, and then I went to work for a startup called Oracle.

When Oracle really was a startup, you were employee No. 20. Right. Turned out to be a pretty good decision. Then, 10 years later, in the early 90s, the information technology business was growing rapidly and we'd automated most aspects of businesses but the processes of customer service and sales were still untouched. In 1993, the state-of-the-art in sales was still a note on the back of a business card. I thought that was highly unlikely to remain the case.

Enter Siebel Systems, which effectively invented the customer relationship management system. That period was such a big step change for tech, in the form of small-factor computing, high-speed relational databases, broadband, and graphical user interfaces, which was huge. We pointed all these technologies at sales and marketing, inventing some things along the way, to create what you know now as the CRM market. We got lucky.

That's an understatement. But that's nothing compared to what's coming now. The next wave is going to be hugely impactful, taking the form of elastic cloud computing, IoT, AI, and big data. It changes everything.

Was there a specific "Eureka" moment which led to the setting up of C3 IoT? I read a lot. I have historical perspective. I know what's going on. There was no epiphany. But I got together with a group of 50 smart people, and we brainstormed this for about a year. I'm a strong believer in the power of collective IQ. To the extent I've been successful, it's because I surround myself with really smart people. And Pat House [Executive Vice Chairman and co-founder of C3 IoT] is one of them; she's a powerful human being. So by 2008, we could see all this coming and decided to set up a company, and a platform, to allow people to deploy, create, and manage AI and IoT projects.

You have a full suite of AI-powered predictive analytic applications now. We began working on this in 2009 and, to date, we've spent approximately $300 million building the platform; 19 billion sensors—each of which is effectively a computer—have been deployed in the last five years, and it'll be 50 billion soon. It's all proving highly scalable and efficient for our customers. We develop applications for them focusing on predictive maintenance, inventory optimization, energy management, and sensor health; we provide a workbench for data scientists; and we utilize a massive enterprise data lake which enables our customers to store data of any size, shape, and speed, and we apply AI and machine learning to all the data ingested.

Let's get specific. The Pentagon's tech innovation hub, DIUx [Defense Innovation Unit Experimental] selected C3 IoT as a partner last November. Can you talk about what you're doing for the armed forces? We won a contract to provide predictive AI maintenance for the United States Air Force—their aircraft assets consist of 5,500 aircraft and 3,900 platforms. The first platform we're taking on is the E-3 AWACS (Sentry) Airborne Warning and Control System. It's an interesting AI problem because there is no telemetry, [recording output from instruments], on what is essentially a Boeing 707 airframe, so it's a natural language processing issue.

If there's no telemetry, what sort of input are you ingesting into the AI system? What we have to work with there are text files. We have maintenance logs, pilot logs, flight records, and so on as sources of data. So we use our systems to go through and look for word pairs, rules, and build out NLP classifiers for the purposes of predicting system failure, and sub-system failure before it happens. I'm not going to share the numbers, although I have them, but a significant number of USAF assets on any given day will not deploy. Understand that these assets are often in very hostile environments and don't have access to the usual backup maintenance facilities. So the idea, for us, is to monitor everything in real time so we know they can deploy.

This isn't monitoring at the device level? No. It's much more comprehensive and intensive than that—it's at system or subsystem level. Airframe, hydraulic, flight controls, avionics. We ingest all this data at the rate they arrive and then run them through an NLP regimen, then an ML [machine learning] process, to identify system failure. In the case of the E-3 AWACS (Sentry) we can do that with about 80 percent precision.

That's huge. We can tell the USAF what's going to fail, when it's going to fail and, most importantly, why. The Pentagon spends approximately 30 to 40 percent of its budget on operations and maintenance. Even a 1 percent gain in efficiency will save billions.

What's the next USAF system you're bringing C3 IoT expertise to?Lockheed C-130 Hercules, the largest plane ever built. It's giant, and they operate most of them out of Travis Air Force Base in Fairfield, California. After that, we go to F-16 Fighting Falcon and F-15 Eagle.

We interviewed the head of AFCEA about the DIUx mission; it's across all armed forces units. Are you just focusing on US Air Force? No. We're about to deploy the C3 IoT Platforms and applications with the United States Navy shortly.

Did the Air Force take you up in an F-16 to say thank you? They haven't provided the opportunity yet, but I wouldn't be surprised.

Let's pivot and talk about the results your platform is getting with its global energy industry clients. A power grid is a large and complex machine, made up of billions of electric meters, transformers, capacitors, power lines, and phasor measurement units. We signed a partnership in June 2016 with Engie, a global energy company, based in France, to build "Engie IoT." Isabelle Kocher, the CEO, is a great visionary and a physicist by background. Engie had revenues of 66 billion Euros in 2016, with customers all over the world in almost 70 countries. We're solving an AI and IoT problem for them, deploying about 100 production applications across all business units—everything from machine to machine, B2C, B2B—with an expected economic benefit of 1.5 billion Euros a year; very aggressive targets.

Can you talk about a specific project? One project we did was for production optimization on combined cycle plants. We modeled two in Bahrain, and were able to identify a 2 percent gap, which is staggering at that level, particularly when you scale that globally. They knew there was a 2 percent difference, but they didn't know why, and using our system, we were able to identify where it was so it could be fixed. It's a very exciting project, taking into account everything from energy efficiency to customer churn. We've set up a center of excellence in Paris, with 12 of our people, and 88 of Engie's employees, working on a very large roadmap that will rapidly deploy applications, every few weeks, over the next few years.

IoT changes the business landscape because it's not about sampling anymore but real time data, right? Right. Previously, due to the cost and complexity of computational capacity and storage, computation was done on sample sets. We'd get some results and use statistics to put confidence levels on those results and do predictions. Now, because of IoT, we have all of the data and all of the signals—it's a complete shift in the way we've done computing for the past 50 years. Predictive analytics is entirely new, with very high levels of precision which was impossible before.

Goodbye, sampling errors. Exactly. As a result, we're looking at a quarter of a trillion dollar software market by 2023: AI, IoT, and big data will affect every industry vertical. This is a staggeringly large software market that we've never seen before. And, in my opinion, the largest sector will be healthcare. The data and telemetry that is becoming available, we can aggregate, and in not so many years time, most people will have their own genome sequenced and be wearing some form of body-based monitors. So we can use AI with very high levels of precision to do predictions for onset of disease.

There'll be no mystery about the signposts on the road to the end [of life] by then. Medicine today, it's about rules-based systems, which is a very blunt instrument. But tomorrow we'll know what's going to happen. We'll be able to take care of people now, with preventative measures, rather than on the surgical operating table later, when the disease is far too advanced. The socio-economic implications of that are huge.

IoT is going to change everything. I think we're really in the AI business now. That's what this is about. I believe we've found ourselves in the right place at the right time.

Catch Tom Siebel at the Code Conference on May 29-31.

This article originally appeared on PCMag.com.