Mapping the Future of AI in Manufacturing with CADDi CEO Yushiro Kato

Watch on YouTube

Summary

At IMTS 2026 in Chicago, SME Media Editor-in-Chief Steve Plumb sits down with CADDi CEO Yushiro Kato to talk about how AI is changing manufacturing. CADDi CEO Yushiro Kato explains how the CADDi AI Data Platform connects disconnected data across engineering, procurement, quality, and maintenance, and why its semantic layer works as a data map that helps AI agents deliver the accuracy, traceability, and repeatability manufacturers need. He walks through products including CADDi Design Review and CADDi Quote, describes how the platform captures tacit knowledge that has never been digitized, and shares CADDi's target of a first business impact within 45 days. He also outlines the company's 10-year vision to accelerate physical innovation by 10x, its $114 million Series D round, its proprietary geometry AI model, and its growing commitment to the US market.

Key takeaways

1. Connected data, one platform. The CADDi AI Data Platform is a vertically integrated AI platform with a data layer, a semantic layer, and an application layer, connecting data from engineering through procurement, quality, and maintenance.
2. AI agents need a map. The semantic layer works as a data map so AI agents can be accurate, traceable, and repeatable, while access controls decide what each person sees, from a contractor to an engineering director.
3. Capturing tacit knowledge. CADDi CEO Yushiro Kato says less than 20% of manufacturing knowledge has been digitized. CADDi Design Review lets teams comment directly on 3D CAD, 2D drawings, and diagrams, capturing why a design was made, not just the result.
4. Speed to value. CADDi aims for a first business impact within 45 days by configuring its off-the-shelf products instead of building systems from scratch.
5. Growth and vision. CADDi's $114 million Series D at a $1.2 billion valuation supports product development, its proprietary geometry AI model, and US expansion, toward a 10-year vision of 10x faster physical innovation.

Chapters

• 00:00 Introduction
• 00:31 About CADDi: connecting disconnected data
• 01:51 The semantic layer and why AI agents need a data map
• 02:32 Product ecosystem: Explorer, Agents, Design Review
• 03:18 Access controls and data governance
• 03:50 Procurement and quoting with CADDi Quote
• 04:44 Capturing tacit and tribal knowledge
• 05:53 First business impact in 45 days
• 06:12 The 10-year vision: 10x physical innovation
• 07:26 Series D and the proprietary geometry AI model
• 08:12 US expansion and connecting data after M&A
• 09:34 Off-the-shelf configuration versus custom builds
• 10:13 Closing

FAQ

What is the CADDi AI Data Platform? As CADDi CEO Yushiro Kato describes it, it is a vertically integrated AI platform for manufacturing. It connects disconnected data, from engineering to procurement to quality, and makes it usable. It has a data layer, a semantic layer, and an application layer of workflow products for design, procurement, quality, and maintenance.

What is a semantic layer, and why does it matter for AI agents? The semantic layer is a data map. CADDi CEO Yushiro Kato explains that without a map, AI agents can only summarize relevant data, which is not accurate enough for manufacturers who need accuracy, traceability, and repeatability. The semantic layer also controls access for each person, so global teams can use data safely.

How does CADDi capture tacit knowledge? CADDi CEO Yushiro Kato says less than 20% of manufacturing knowledge has been digitized. CADDi Design Review lets teams comment directly on 3D CAD, 2D drawings, and diagrams as part of their workflow, which captures context data, such as why a design was made a certain way and what mistakes were found.

How quickly can manufacturers see results? CADDi targets the very first business impact within 45 days, starting with a smaller scope and then building additional impacts. Because the products are off-the-shelf and highly configurable, CADDi configures them rather than building systems from scratch.

What is CADDi's 10-year vision? CADDi's vision is to accelerate physical innovation by 10x, cutting the time it takes to make physical things to one tenth. To get there, CADDi announced a new generation operating system for manufacturing where customers can leverage and generate data.

Related links

• CADDi AI Data Platform
• CADDi Design Review
• CADDi Quote
• CADDi Explorer
• More CADDi news

Full transcript

Source: SME Media, YouTube, published September 25, 2026 (https://www.youtube.com/watch?v=AJ5S1UBtI6U) Recorded at: IMTS 2026, Chicago Speakers: Steve Plumb, Editor-in-Chief, SME Media (interviewer); Yushiro Kato, Co-Founder and CEO, CADDi (guest)

Editor's note: Cleaned from YouTube's auto-generated captions. Filler words (uh, um) and repeated words are removed. Speech-recognition errors for company and product names are corrected (for example "Katana," "Caddy," and "Catia" to "CADDi"). Passages the captions could not resolve are marked [inaudible]. Square brackets mark editorial insertions. Timestamps are MM:SS from the captions.

[00:00] Yushiro Kato (opening clip): Data from different companies should be connected, but it's not connected. So we help our customers to connect disconnected systems and help them create synergies across companies.

[00:18] Steve Plumb: I am Steve Plumb, editor-in-chief of SME Media. Here today with Yushiro Kato, co-founder and CEO of CADDi.

Yushiro Kato: Yes.

[00:26] Steve Plumb: Thanks for joining us. We're at IMTS 2026 in Chicago.

Yushiro Kato: Yeah, thank you for having me.

[00:31] Steve Plumb: And you had some big news this week and last month, but before we get into that, tell us a little bit about the company. It's changed quite a bit since its founding in 2017, and your background: you were on the consulting side and then jumped into the manufacturing side.

[00:49] Yushiro Kato: Yeah, so long story short, we are CADDi. We provide an AI data platform for manufacturing. It's a vertically integrated AI platform for manufacturing. So we basically enable our customers to connect the disconnected data, from engineering to procurement to quality defects, and then make it usable. Our application layer can also generate and capture tacit knowledge that has never been captured, and then we create a data flywheel for manufacturing so that they can leverage data, basically, like [inaudible] knowledge.

[01:23] Steve Plumb: So you have a number of different products, and you keep expanding, and the digital platform kind of encompasses everything.

[01:31] Yushiro Kato: Yep. Our platform is vertically integrated. So we have a data layer and a semantic layer, all the way to an application layer with six vertical products, from products for conceptual design and detail design all the way to products for procurement, quality, and maintenance.

[01:51] Steve Plumb: So, you noted the data layer as well. What is the semantic layer?

[01:55] Yushiro Kato: Yeah, so the semantic layer is a data map. Of course, in this era, you need to be able to leverage AI agents. And AI agents need a map, like human beings. Without a map, it can only summarize relevant data, but it's not enough, or it's not accurate enough, for manufacturing people, because manufacturing companies need accuracy, traceability, repeatability. So our semantic layer enables our customers and the AI agents to basically work on the data more correctly, and repeatedly, basically.

[02:32] Steve Plumb: So how many different tools are there that feed into the digital platform area? You had the [CADDi AI Data Platform] that was introduced 3 or 4 years ago, and now Explorer. What all is there?

[02:43] Yushiro Kato: Yeah, so we have two products on what we call the discovery and agent layer: CADDi Explorer and CADDi Agents. They are basically the products that enable our customers to search and discover data and leverage data. On top of that, we have the workflow product layer. We have six products, like CADDi Design Review, which is a design review platform. AI automatically checks the design drawings and 3D CADs and then identifies mistakes and errors on drawings so that they can mitigate the risk [inaudible]. There are six products.

[03:18] Steve Plumb: Yeah. And basically anybody in the organization can access this information?

[03:23] Yushiro Kato: Yes. Another point about the semantic layer is that it has access controls. So we have a lot of global enterprise companies, and of course, contractors in Mexico and, I don't know, engineering directors in the US should have different authorities, right? So our semantic layer enables our customers to control the access for each and every person, so that they can safely use and leverage the data.

[03:50] Steve Plumb: So a lot of the AI applications are starting off on the quoting side, something like the CAD/CAM drawings. Is that where you're starting off as well?

[04:01] Yushiro Kato: Yeah. So we actually started our company with a business called CADDi Manufacturing, which is an Amazon for custom-made parts.

Steve Plumb: Yeah.

[04:10] Yushiro Kato: And so we work with a lot of procurement officers in the US. So we started with a business called CADDi Quote on top of the data platform. CADDi Quote is basically an automatic quotation platform. It can automatically select suppliers based on similar parts in the past, and it can also give you a benchmark on the cost of the products.

Steve Plumb: Mhm.

[04:34] Yushiro Kato: So it streamlines the procurement processes and then reduces the cost. So that's one of the core products on the workflow product layer.

[04:44] Steve Plumb: And you mentioned tacit information. How does that differ from existing tribal knowledge? What exactly is tacit information?

[04:53] Yushiro Kato: So, unfortunately, less than 20% of the knowledge in manufacturing companies has been digitized.

Steve Plumb: Mhm.

[05:02] Yushiro Kato: So more than 80% is tacit knowledge or tribal knowledge in human brains, right? So you need to capture it in order to fully leverage your assets as an organization. So what we do, for example: our CADDi Design Review platform enables our customers to put comments on the design data itself, like 3D CADs, 2D drawings, diagrams, etc. And it's built into the workflow, and it automatically captures new data, like the context of the [design]. Manufacturing companies have result data, like drawings, 3D CADs, designs, but they do not have context data. Why is this design designed this way? What are the mistakes found in the workflow? That's missing. So we automatically capture that with our [inaudible].

[05:49] Steve Plumb: [inaudible] that type of information and build that out.

[05:53] Yushiro Kato: The quickness is super important. So we basically target 45 days to create the very first business impact, and then build another impact, and another impact, and another impact. So we should focus on a smaller scope, but the timeline is there.

[06:12] Steve Plumb: I caught a little bit of your presentation yesterday. You were talking about a 10-year plan, 2035. Is it possible to think that far out? And what are some of the things you're looking at in that time frame?

[06:27] Yushiro Kato: Yeah, so we announced a 10-year vision with a visual demo. In 10 years, our vision is to accelerate physical innovation by 10x. Everybody's talking about AI, and AI is creating new theories and new ideas exponentially. But when we look at the physical world we're living in, how much has it changed, right? There are not a lot of changes in commutes, hospitals, offices, houses, whatever. We live in the physical world, so someone needs to change the physical things, and someone needs to make physical things. But the time frame to make something is very long. So our vision is to shorten it, to cut it to one tenth. And in order to do that, we announced the new generation operating system for manufacturing where, again, our customers can leverage data and generate data.

[07:26] Steve Plumb: Yeah. And just last week you announced the latest investment round, a Series D, I believe. Tell us a little bit about that, 100-plus million, and what that means to the company in terms of future investment.

[07:41] Yushiro Kato: Yeah, absolutely. So we announced the Series D round with $114 million at a valuation of $1.2 billion. With this funding, we want to boost, of course, product development. And also, we actually own our proprietary AI model to analyze the shape of parts, the geometry of parts, that is super specific to the manufacturing industry. So we want to deepen the technology of our AI model. And also, we want to expand in the US a lot more.

[08:12] Steve Plumb: Right. So this is about my next point, the expansion in the US. Obviously, we're here at IMTS, an important manufacturing show. What does that process look like as far as the expansion?

[08:25] Yushiro Kato: Yeah, so we came to the US 4 years ago. I personally live in Chicago now. We are fully committed to the US market. Actually, the issues that US companies have and Japanese companies have are almost the same. They do have, of course, [inaudible], like systems of record. They run factories with similar operations. And now, after 4 years, our product is used by a lot of large enterprise companies in the US, which is great. But from now on, I don't know if we want to localize the products, because the issues are the same globally. But in the US, there are a lot of M&As, right? Companies buy companies. So the data structure is relatively messy, because data from different companies should be connected, but it's not connected. So we help our customers to connect disconnected systems and help them create synergies across companies.

[09:34] Steve Plumb: How much is customized versus, you know, can be off-the-shelf?

[09:39] Yushiro Kato: So we have, again, the data layer and semantic layer, and also eight products on top of that.

Steve Plumb: Mhm.

[09:45] Yushiro Kato: All of them are configurable, highly configurable, but we do not create systems from scratch for each customer, because speed of implementation is critical. And we know manufacturing. We have, of course, a lot of manufacturing customers.

Steve Plumb: Yeah.

[10:01] Yushiro Kato: So we know how their semantic layer should be designed, for example. So we basically leverage our off-the-shelf products and customize and configure the top 20% of the products, basically.

[10:13] Steve Plumb: All right. Well, congratulations again. Look forward to seeing what you do next.

Yushiro Kato: Thank you.