Arundo Analytics

TOR JAKOB RAMSAY: So my name is Tor Jakob Ramsay. I'm the founder and the CEO of Arundo. We've been in business now for 2 and 1/2 years, so I got an idea-- I was a three to four years ago. I was for 15 years a senior partner in McKinsey. And I was leading actually technology practice globally in asset heavy industries.

And of course, like every consulting company in the world, you were doing some work of figuring out what's the industry 4.0 and that going to be about, and figured out that the old asset heavy industries, they a lot of data, but they're not using it in any kind of a systematic way, or storing it historically. And so the idea was a little bit to bring what's happening in more consumer oriented industries.

How could the Amazon recommend a book to buy, or Netflix what movie you want to see. How is American Express doing fraud detection. Could we apply some of these algorithms, those techniques of identifying a failure before a failure is happening on an oil rig. Could we optimize production in an aluminum mill or something like that by understanding when is good and when is bad production.

So I left McKinsey. I founded Arundo. And we actually build the company. We did it actually pretty different. We did it on three locations at the same time. We did it in Silicon Valley, we did it in Houston, because the biggest oil and gas market which is our prime segment initially is based in Houston.

But some of the most advanced applications are in the north sea. So we also have an office in Oslo. Currently of course we also have an office now in Boston. So that's a little bit the origin of the company. So the whole idea is to take all the data [? that ?] already exist historically. We have built technology to capture and analyze live streaming analytics, and based on that provide solutions that could optimize your operations.

So our main focus has been on three industry verticals. Oil and gas, shipping, and renewables. Initially it was mostly oil and gas. The reason is because they literally invented big data 30, 40 years ago in the seismic modeling and geophysics. So they have a lot of data. They have very advanced equipment. It's a lot of sensors. It's a lot of data to access. So there's no problem of getting access to data.

And it's a big industry. It's the biggest industry in the world. It's very similar when you move into other segments, which is shipping. It's the same equipment manufacturers, and it's the same challenges also in utilities. So it's all about increased asset utilization. Which means you need to understand how you could be in [INAUDIBLE] time, and how we could increase productivity with the same level, or less resources, safer and better.

So if that was our three initial segments, we also now have customers in mining. We also have people or companies in discrete manufacturing. So all asset heavy industries are over customers. Even also companies, who is providing equipment to those industries.

So one of our initial customers, or that we build a lot of technology, was [INAUDIBLE] is also an MIT ILP company, so it's quite fun that we are also part of the MIT STEX and ILP. One of the problem they wanted to have help with was how can we understand what is driving failures on the compressor. So a compressor is typically, if you're producing gas, you need a compressor to get all the gas pushed through continental Europe to Germany as an example. So if that one is failing, you're losing your production.

So what we did was build algorithms, build all also the technology to capture and analyze this data so we could figure out what was explaining the failure. And then when you have a historical model, you apply that one on the real data, real time streaming. And then you can do predictive operations, or predictive maintenance, and avoid failures.

So you could change your production profile. Or if the part is going to break anyhow, you could then have a controlled stop, rather than an unplanned stop. So that was actually one of our first use cases, and one of our first customers we had.

TOR JAKOB RAMSOY: So what our company has done that makes us unique, we have focused very much on the biggest problems in operational analytics and how could you get machine learning into your daily operations. So the first problem is a lot of data science today is dying in PowerPoint. So if you don't manage to deploy it out on the assets-- if that's a production plant in China, if it's an oil platform in the Mexican Gulf, if it's an oil platform in the North Sea-- we have built a tool that you could deploy your machine learning model, upload it on the cloud, get it out on the asset within seconds. That's the first problem.

The second problem-- in the future, we think most industrial companies will have several machine learning artificial intelligence model. How are you going to manage all those models? How are you going to make sure that the data is getting in, that the model is giving the quality that you want to have, that the right people have access to it? So think about DevOps for machine learning. We call that model management. To our knowledge, we are the only company who has a solution for this, because our belief and philosophy is that industrial internet is going to be an open internet.

And then, the third and the unique part we have is what we call edge analytics. So we have created an edge agent, which is doing two things. One is to gather and standardize the data on the edge. Probably that's not the most unique, and you could find other companies who've done that. What's really unique is that a way you can push your models out to the edge. So you can run the analytics on the edge, so that means you don't need to transmit all your data to a cloud-based model. You do it on the edge. And then the only thing you need to communicate is the model changes.

And so every time you have a compressor somewhere in the world who is seeing something new, and then you have the thousands other compressors who could benefit from that, you communicate that model change. So then you have the wisdom of the crowd. And that's a technology we also are very unique.

Another unique customer situation we have-- or it's actually a partner and an investor of ours is a Swiss-based company called SICPA. They are a 90-year company, so literally every note or every currency in the world except North Korean wons and Japanese yen are produced by them. So they've been around and creating what's called secure markers. They reached out to us and created a partnership with us for how could we do something similar in the oil and gas industry.

So how could you use chemical markers from the well head to the depot, from the depot to the gas station, and make sure that you have track and traceability of your hydrocarbon flows? And we are providing the digital and analytic solutions for that, which is the global markets, and they are uniquely positioned, have unique relationship with it.

[MUSIC PLAYING]

TOR JAKOB RAMSEY: To explain a little bit about over uniqueness, and capabilities, and why we can do it now. And we probably couldn't have done this three or four years ago. It's because of all industrial equipment are now connected with sensors, but also, that you have cloud computing. Which is safe, and it's cheap.

So one of our customers, they're one of the biggest producers of industrial tread. They have a plant in China, and we could actually put that plant online within 90 days. So within 90 days from we start working with our customers, we are capable of delivering business value. Which means we could stream data from a plant in China. We could provide analytics from the US, from UK, or anywhere they want to have it in the world.

We could set and deploy the models. Our customers ideally are using our technology, building new machine learning artificial intelligence models, improving production in China, or wherever they have a plant in the world. And that's done with one click and within very limited resources. So we are really looking forward. I'm very proud to be part of the MIT STEX25. We see that as a unique resource of getting access to a lot of the industrial companies who's part of the greater MIT ecosystem.

Another thing is the research, and all the faculty, and the students here. Students also are potentially recruits, because we want to tap into the fantastic talent pool at MIT. So that's the three things we are really, really looking forward to.

What really makes me excited about Arundo is all the opportunities we still not have tapped into, so think about all industrial equipment in the world. Think about the pump. Think about a compressor, so no one in the future wants to buy a pump. They want to buy it pumping, so you're going from that product based to a service based economy.

When you do that, you need more data, because you need to understand what is delivering, how it's working. So we see a huge market in all the equipment manufacturers in the world or all that OEMs in the world. All of them would need a digital operating system. So they could connect their pump, their compressors, turbines, or whatever it is into the cloud, be able to provide analytical services, and to sell it based on output rather than actually on input.

And all these manufacturers are owned in the world. Where should it go? It doesn't exist to some kind of-- It doesn't exist in Microsoft Windows or Google Android. So we think about us as the Android of the industrial internet. That is really what makes me excited.

TOR JAKOB RAMSEY: When we meet our customers, because the industrial internet is a lot of people is talking about it. But there's very few who have delivered industry scale production ready solutions who is really working and delivering business value. So one of the unique things that we do, and how we deliver it, we set set at within 90 days, we guarantee a delivery of business value, 90 days.

So that's not the slide work, but it's an online. It's a real model. It's a real reproduction solution that is working and help our customers to improve their production. The way we do it, it's cloud based, it's models out of the box, and we could connect to literally all the industrial equipment to a system.

If it doesn't have any kind of a way of streaming, we have also made a technology partnership with Cisco and Dell. So we could provide a cloud computing and a edge computing environment. All this we do in 90 days.

If a company managed to capture the fantastic value it is by understanding equipment behavior and interaction man machine, which is also part of what we could fix. So think about an oil rig in the North Sea. The average utilization is actually 84%. The planned uptime is 95. So then you have an 11 percentage point downtime, which is not planned.

The value of that if it's an oil field who's producing 50,000 barrels a day with an oil price of 45, that's $100 million US in lost revenues. So if you then gather the data, you manage to understand what this data is explaining. And you fix that uptime, downtime, which could be human. It could be machine. Usually, it's a combination of both.

That is what you can do with machine learning. That is what you can do with our technology to capture that unplanned downtime or increase your asset utilization. So if a company has 50 oil rigs as an example, that's $5 billion in lost revenues. So the magnitude of the business cases are enormous, and industrial internet, I still think we are talking about year one even if people have been talking about it for a long time.

[MUSIC PLAYING]