Left: Hiroki Kamobayashi, CEO, Arent, Inc.
Right: Seitaro Narue, ex-Piping Engineer, Chiyoda Corporation & now-CEO, PlantStream Inc.
The two of you, Seitaro and Hiroki, have crossed over organizational boundaries to tackle problems together. How did you originally meet?
Seitaro (Chiyoda): It was back when Chiyoda Corporation was pursuing various AI start ups and proofs of concept regarding the automation of piping design. I met Hiroki on May 10, 2018.
Hiroki (Arent): You remember the exact day!? (Laughs)
Seitaro: Of course. After all, when we first met and discussed automation, you said something that shocked me. “We should not use AI. We should rather build a specialized CAD system.” We had asked a lot of technical questions of Arent’s engineers. Their answers made us realize that they were a good match for us. That was the catalyst for us to start development in collaboration with Arent.
Hiroki: I had brought along a member of our staff who was highly proficient in CAD to do the talking. This engineer dedicated himself to learning about plant engineering. Over the course of our discussions with Seitaro, our direction became extremely clear.
Seitaro: Those two hours were so informative. I was already confident we could do this. All I had to do was figure out how to explain it internally at Chiyoda. AI was booming, and internally we had already been thinking in terms of AI. Now it was like, “Can we really pull off CAD development starting from scratch?” I had to be persistent in using Arent’s technology to dispel everyone’s fears, one by one.
Hiroki: What really got everyone on board was submitting a proof of concept, developed over two or three months, to show we really could do this. We then began slowly broadening the scope. It was crucial that we provide concrete proof of the scale and quality of what we could do.
Seitaro: We started really small but kept moving forward as we produced results.
Hiroki: Seitaro presented what we produced to his superiors at Chiyoda, securing approval for the budget. From there, it gradually became a larger refined project. That is how I see using agile successfully.
Once the project began, where did you start from?
Hiroki: We knew this would become a pretty large-scale project, so we started from team building. We already had several engineers who were good at CAD, but we needed as many as possible. Right now, we have about 50 people.
Seitaro: As a manager, weren’t you worried about going over budget, though? (Laughs)
Hiroki: Well, I was counting on Chiyoda for securing the budget for us. (Laughs) I knew how serious Seitaro was about this. He told me that, so long as we showed off a good product each month, he’d ensure it kept going. So I decided to make sure the wheels kept turning.
Seitaro: Well, I had no choice but to say I could make it happen. (Laughs) Of course, internally, a lot of problems came up. But I never saw them as a hurdle that couldn’t be overcome.
As the product gradually came together, what was the internal company response like?
Seitaro: In a way, the demands have gotten even stronger than I expected. (Laughs) People want to make it even more automated. You only get that kind of response if you have a good product. It made me realize how important it is to show off the real thing. Around eight months after the start of the project, we showed off the progress and secured a supplementary budget, and began considering turning it into a commercial product.
Hiroki: At the start, everyone was pretty apathetic to our proposal. When we first showed the product, we got a lot of criticism. “This is no good,” “We can’t deploy this production,” ”You’re not using AI yet,” and so on. Of course, at that point, it wasn’t in a deployable state, so we’d already taken that criticism into account. What really mattered was getting people’s opinions about the product’s end goal, and whether they’d react by starting to get other people on board.
Seitaro: We had several internal evaluation meetings, and we had a bigger audience every time. The first development report had 5 or 6 people, the next 30, and the last one 100. Even higher-ups and design department personnel. As we went on, people had more and more to say, and we were able to get input from a variety of points of view.
Hiroki: Since the product was well put together, we wanted people to take a look and really get into it. Positive opinions are great, but we appreciate criticism, too.
Seitaro: Chiyoda was most impressed with the fact that the Arent engineers’ explanations could’ve been confusing for a plant technician. The fact that everyone was so well-studied really got across how prepared Arent was. I think a lot of the technicians felt the same way.
It was mentioned you still aren’t using AI. Why is that?
Hiroki: We lack the big data required, so we can’t be confident in the accuracy of AI. Of course, there’s always a possibility that we’ll introduce AI in future, but at present, we’re focused on a more standard implementation. Even without AI, we can produce clear, mathematically-expressible logic by gathering information from plant engineers. At a glance, it seems like an AI-solvable problem. So we started out thinking we would use AI while creating this project. However, plants are extremely complex and studying every nuance of an individual plant was too big of a prerequisite.
Seitaro: Your average AI engineer tended to assume that they could perform numeric analysis on the data they received and produce a product based on that alone. In contrast, Arent worked toward a comprehensive grasp of the subject and its problems, right down to the finer details.
Unleash the expert’s potential.
From there, you began to create a joint venture, PlantStream Inc., and moved to sell the system externally, which wasn’t in the original scope. What was the cause of that?
Hiroki: From Arent’s point of view, it seemed a waste to stop development of this product midway through. System development is fundamentally about repeated, long-term revision to achieve a competitive advantage. If the system wasn’t sold externally, it would be “complete,” development would slow, and some day the system would be surpassed by a maintained one. I felt that external sales were essential to ensure the continued strength of the product. Still, marketing consumes a great deal of energy totally separate from development, so it’s a big risk. This is where the idea for forming a joint venture came from, so Arent would be sharing the same risk. That would allow development to continue unabated, and if the project was a success, Seitaro’s team would be our partners in that success. That’s why I proposed the idea.
Seitaro: As Hiroki said, we wanted to be in the same boat. With the fusion of Arent’s CAD technology and our plant expertise, we were prepared to proceed together as 50:50 partners. Personally, I’d had my eye on external sales from the beginning. But it wasn’t at all within the company’s scope, internally. Since it was like saying, “Let’s turn our strengths into a system and sell it to our competitors!” But, as I said, the product turned those opinions into approval, and finally we decided to commercialize it.
Hiroki: Once we had the product itself at the center of things, the more complete it became, the more people got on board. Even those who were opposed initially eventually understood. Once we showed it was practical and feasible, people began to see the value of it.
How does this change the landscape?
Seitaro: I think the most groundbreaking aspect of this project is the new forms of value it derives from plant engineers. Their purpose has always been to produce profit through their design work. We’ve created a new way to capture the specialist engineer’s knowhow.
Hiroki: We have to do a better job of sharing such know-how and knowledge. That applies to the industry as a whole.
And one example of such an industry is plant engineering?
Hiroki: That’s right. So now that we’re doing this, we want to go all the way. We want to change plant engineering itself, everything from upstream plant design to downstream construction.
Seitaro: We see this as a product that will revolutionize existing workflows. First for Chiyoda, then spreading to many other plant builders.
Hiroki: Plants themselves exist to transform resources into more valuable forms. The conversion of naphtha into gasoline, for example. Similarly, with PlantStream, we aim to change the plants themselves, as well as the worldwide flow of production itself. We like to dream big.