Preserving Tribal Knowledge with AI — Engineering's Hidden Asset

Episode Summary

The episode titled "Preserving Tribal Knowledge with AI — Engineering's Hidden Asset" explores how artificial intelligence can be leveraged to manage and utilize tribal knowledge within complex engineering environments. The discussion panel includes Ye Wang, CEO of EverCurrent, a startup focused on developing an AI-native platform for managing product data; and Rob Ferroni, known for his insights into Product Lifecycle Management (PLM) and the impact of AI in manufacturing processes. Both guests share their experiences from leading cutting-edge software development teams and emphasize the importance of tribal knowledge management in leveraging AI effectively.

Key technical insights discussed include the challenges large engineering teams face in managing and utilizing their extensive tribal knowledge, particularly when integrating AI technologies. Ye Wang highlights that while larger teams have more resources, they often struggle with how to harness this knowledge for AI-driven benefits. Rob Ferroni underscores the need for rethinking work processes to fully integrate AI, suggesting that companies achieving significant breakthroughs will be those that fundamentally change their approach to work.

For PLM and engineering professionals, the key takeaway is recognizing the potential of AI in preserving and utilizing tribal knowledge. By adopting an AI-native platform, teams can better manage complex data dependencies and track the impact of changes, ultimately enhancing operational efficiency and innovation.

---

Full Transcript

Michael Finocchiaro

And I believe we're live. This is Michael Finicaro on the AI Across the Product Lifecycle podcast. Ugly Christmas sweater edition. We're going to talk today about tribal knowledge in AI. And I'm really excited to have Rob Ferroni, one of my regular panelists on the ⁓ Future BLM podcast, as well as Ye Wang, someone I've met recently from a startup called EverCurrent. who has made it her mission to work on tribal knowledge. So I'll let maybe Ye introduce herself and then Rob.

Ye Wang (CEO Of EverCurrent)

Awesome. Yeah. So glad to be on a podcast. So I'm calling from San Francisco. So it's early morning right here. ⁓ yes. Yeah. We're next. We're actually Oakland. So next to the ocean, like the Bay. Yeah. That's a nice spot.

Michael Finocchiaro

We can see the birds flying behind you sometimes, you know. I miss the A's already, even though I don't live there, but okay.

Ye Wang (CEO Of EverCurrent)

Yeah, awesome. my background, I really came from a computer science background, but all my career has been working on the bleeding edge of building software for hardware. So I started my career doing more 3D printing to watch optimization, like the cool stuff, more than a decade ago at MIT and Autodesk. I think the lesson I learned from that experience was fast iteration really brings innovation for hardware teams. After that, I joined Onshape when it was small. It's a very MIT-heavy team in Boston. And it was a time WebGL just came out and was so impressive when I first saw that demo. It was like, wow, you can build an engine in a browser. I was like, OK, this is a team I'm going to join. And I think now it's picking up a lot of speed and it's becoming more and more mainstream. I actually see large teams using Onshape. After that, started my first startup in construction tech.

Michael Finocchiaro

Yep.

Ye Wang (CEO Of EverCurrent)

I'll skip a little bit. And then the last several years, I've been leading generative AI research for automotive at Autodesk. I think that's where I encounter a lot of tribal knowledge. I think before, I work with teams of different size, hardware development team, small, medium, large. But when I started working with large teams, realized not only, I mean, not just because you have more resource, you have your tribal knowledge together.

Michael Finocchiaro

Okay.

Ye Wang (CEO Of EverCurrent)

Sometimes some of the teams have even more issues managing their tribal knowledge. And they have so much, they're sitting on a well, but they don't know how to use it. And when we came in, trying to bring AI technology, I think the biggest blocker is how do you use your tribal knowledge so you can use AI in a way that's actually beneficial for your operation. Anyway, so that's the whole inspiration of starting EverCurrent. So EverCurrent is an AI native platform. really trying to get people's knowledge together, getting the dependency together, getting people to understand the impact of change, to track the impact of change.

Michael Finocchiaro

You found it this year, right? This is a pretty new venture for you. That's exciting.

Ye Wang (CEO Of EverCurrent)

That's right. I found it this year. Yeah, yeah. And we're backed by Lacy's Moosey's Be Wrong and many other amazing investors.

Michael Finocchiaro

That's awesome. ⁓ Rob, you and I have got way back. So I know that that's one of your sticks is tribal knowledge. And of course, you're welcome to pitch a share PLM's conference coming up because they will be talking about that there too, right? In a couple of months.

Rob Ferrone

Hmm. Yeah. Yeah. think maybe something that people don't realize about me or don't know about me because I'm often reacting to PLM content or creating PLM content. I've adopted the term PLM ⁓ for managing product data across the life cycle within complex engineering environments. But actually, ⁓ I never really interacted much with PLM systems. I was far more involved on the human side, the process side, the actual information flow and the business side. So, you know, I've been working in engineering environments for 25 years and I've been making PLM work. I've been bringing PLM to life and I started a company which had an army of people which brought PLM to life. And so I really understand the context around engineering, manufacturing and service and maintenance. In terms of getting the right information in the right format to the right people at the right time in order to drive the right outcomes. so I think when we have conversations around PLM, so much of the conversation is focused on technology and even if we're not talking about a PLM system, but a system architecture, it's technology based. But that is almost ignoring 90 % of what makes PLM as a business strategy come to life. And that is the people involved, the conversations, the meetings, where things attract even PMO. And so that's why I'm really excited to be here today and to be able to get into what Yi is doing and what she's thinking around what our product is going to do. because I had an army of 400 people doing this kind of work and I think that actually what Yi is proposing could be the next generation of that type of work.

Michael Finocchiaro

Very cool. Well, some of the people on the call may not quite understand this term, tribal knowledge, or maybe we ought to pose it and put some boundaries around it so we know really what we're talking about. I don't know who wants to field that one, but how would you guys define tribal knowledge with respect to engineering and manufacturing, of course?

Ye Wang (CEO Of EverCurrent)

I'm curious about Rob's definition. think everyone has a different definition. I will just quickly say what I think this is. I think this is a term, say you're an aerospace company, You're an aerospace company and then you hire the most talented engineers. A lot of engineers, my work don't say Apple, right? But then you come into like working in aerospace.

Michael Finocchiaro

Alright, go for it Rob.

Ye Wang (CEO Of EverCurrent)

Then they would say, ⁓ there's a lot of tribal knowledge we need to pass to these new people. Super talented engineers, but you just have never developed aeroplanes before. And I think the reason they say tribal knowledge is maybe also because it's only existing tribal memory. Like it's only in some people's heads of how we do things, right? And what are the dead paths to avoid? And it has been really difficult to document. That's my take. I'm curious what Ralph thinks.

Michael Finocchiaro

Right.

Rob Ferrone

Yeah, I agree. I think it's the culture, it's the habits and the stuff built into people that been in the organization and the industry for a long time. I'll give you a couple of concrete examples. know an automotive startup company that is referring to standards. automotive standards. And they're very keen on having that with the with their suppliers. Whereas a lot of the established automotive companies, they've, they've gone beyond that, because they actually have their own standards, their own kind of tribal ways of doing things. And so they don't actually reference the, you know, the standards anymore. Yeah, and I think the interesting thing about this is, is, you know, it's pros and cons, so could be super valuable in the sense that everyone just knows instinctively what they need to do and the right way of doing things. But on the other hand, you know, you have this, this is the way we've always done it. And actually, you know, that that holds people back sometimes. And that's why a lot of companies are now saying, you know, what else is out there? What what insight can you bring from the outside world in? And, and I think there's a lot of companies at the moment as well that are shedding workforce and ⁓ you know, I've seen firsthand the, the issues caused by that where tribal knowledge is leaving the building. ⁓ and, ⁓ you know, companies don't know how to make decisions anymore. They're, they're, you know, they're going back to basic problems that were, ⁓ fixed in the past because people just instinctively knew. Go to this system, go to this place, speak to that person. ⁓ no, we can't do that. Or yes, we do it in this way. Whereas it's like people having to. Go through that learning curve all again

Michael Finocchiaro

So maybe the, I suppose that there could be also a bit of entropy because if people are not using the system, it's just gonna get worse. ⁓ And if people are, well, if you implemented the PLM but you didn't take the people into account in the first place, then it's got a kind of in a bad standing and then it just gets worse over time as people talk about it and the tribal knowledge kind of erodes the confidence in the schools. ⁓ I guess Everkern was designed to sort of correct that direction, right? To kind of get things back into the straight and narrow.

Ye Wang (CEO Of EverCurrent)

I we're not trying to correct anything. I think we plug into a lot of market movement. I was reading this report yesterday, If you look at ⁓ VC money, like earning early seed, there are a lot more money going into hardware these days. I think what it reflects is there are a lot of innovation that people are actually really excited about hardware these days. What are the new innovation of like physical things that people can bring to the world to change the world in very fundamental ways. I think the other part is development speed. People need a lot more development speed. There are traditional companies, maybe their cycle, still using very waterfall kind of process. Their cycle might be four years from planning to actually shipping the product.

Michael Finocchiaro

Hmm.

Ye Wang (CEO Of EverCurrent)

you know, does that help them to be stay very competitive in this market? Maybe not. Maybe it needs to be three years. Maybe two years, maybe one year. Right. But I think like reflecting back what Rob was saying, it was really interesting. It's like, there is also some muscle memory in how you've been doing things. For organizations that always forget their tribal knowledge, they have issues, they run into bigger risks. For companies, you know, that understand and remember their tribal knowledge so well and follow you so well. they might have a little bit like speed issue or innovation issue. So anyway, coming back, think fundamentally like hardware, people are really busy. You're always like racing towards releasing, then you're starting a new program. There's very little time for you to sit back to be like, okay, I think this time we can do this program totally differently and let me play out the scenarios. Like there's no time for people to do that kind of thing.

Michael Finocchiaro

Hmm.

Ye Wang (CEO Of EverCurrent)

you're always putting off fire when things are happening, when your PCB is stuck in the custom. You're sitting there figuring that out instead of thinking like fundamentally, how can the development cycle be different? What are the new options there? Anyway, so coming back to that, I think ever-curing is really like, first of all, you know, getting your tribal knowledge together so you can reuse some of the knowledge, you can onboard people faster. Like I think the basic stuff, but the bigger thing is like, Let AI to in the background to help certain things run a little bit more smoothly. You'll release a little bit more time, more capacity for engineering team to focus the innovation that you want to focus on. Yeah, so that's kind of like my perspective around it. Yeah.

Michael Finocchiaro

right?

Rob Ferrone

Yeah. I mean, there's a couple of things to bring into this conversation. One of them was actually, ⁓ Oleg mentioned in a post yesterday, we talked about, ⁓ he talked about AI automation and agents. he was saying that, ⁓ rightly so as well, you can't just have AI agents automating the processes of today because, you know, the real advantage is to actually change completely the way that work is being done. So you're not looking to kind of replace the activities that people are doing, but rethink it. So completely deconstruct it. And if you look at any engineering operation, know, manufacturing operations and the process of product innovation, and really deconstructed that work, you know, so what time is being spent looking for information? What time is being spent communicating? What time is being spent thinking, you know? And if you, you know, take the tasks and deconstruct that all out. then you'll be able to see, okay, why is it being done like that? Okay, it's being done like that, because that's the way it's evolved. There's a human involved, you know, so it has to be done like this, because that's the human in human human interface. But actually, if we if we looked at this with fresh eyes, we would do this completely differently. You know, and I think that's, that's the really interesting space here, and the opportunity. And that's why I think, again, that that what what Yee's doing is got, this could be the way to actually create the leap in PLM and engineering ops that we've not necessarily seen with systems alone. So I think that's for me the kind of, yeah, exciting space.

Ye Wang (CEO Of EverCurrent)

I want to add to that a little bit if possible. I was just chatting with some engineer friends yesterday. We're talking about the complexity tax engineers pay. Engineering teams pay. And that basically can be up to, like, say, one third of your engineering capacity. I think people just didn't realize that. What's our complexity tax? Basically, Rob mentioned it's like,

Michael Finocchiaro

So is this? ⁓

Ye Wang (CEO Of EverCurrent)

Gathering context, figuring out. Some of don't realize engineers spend a lot of time gathering context, understanding whose change affects me, who I need to talk to. Documentation. This thing has happened. Other people need to know. And the ripple effects, the more complex system you build, the ripple effects are just harder for anyone to know, to understand. And those are the things, I think traditionally, are managed in hard-coded ways. If you have a very good PLM system, you follow those steps. But then the process, then what happens? The process can be long. It takes a bunch of people to review it. And then people start to have side process that they do not get into their change management as often. I think the way we look at how we fit in is You your side process also need a little bit understanding of the dependency of the change, right? And then let's first help you get through that. But it's very, very important to understand what are the critical dependencies and how you propagate it through. And that is AI system, better job for AI system than for any individual to kind of try to piece it together, spending so much time piecing that together.

Michael Finocchiaro

you

Rob Ferrone

You know, we had this podcast the other day, didn't we, the 737 Max. And Patrick loves talking about complexity and he argues that actually what we need to be doing is focused on reducing complexity. But I'm totally with you in the sense that when we used to sit in a room, when teams used to sit in a room and they could communicate directly, know, every, you had very low friction and then you started to have departments, functions, silos, people sitting in different buildings. had different teams. had

Michael Finocchiaro

Yep.

Rob Ferrone

you then had to create the systems and the PMO to actually, you know, coordinate people and integrate people. And that's, yeah, added a huge layer as an overhead, you know, for your business. then you've, so you've got, you know, within a big OEM, big companies, have, you know, people who are their role in that business is integration between teams, it's to do the, you know, the processes in the background in parallel. just to make the core stuff work. know, saying that is interesting because in so in many companies, they're bloated and they have, you know, too much headcount making the business work, being the oil in the gearbox. And in the startups, it's I always, the parallel I draw is like an orchestra coming together that have never played before. They've got lots of talented musicians, but, you know, they're writing the music as they're playing it and they're trying to, you know, get a feeling of how to work together.

Ye Wang (CEO Of EverCurrent)

I'm filming it. I'm filming it. I'm doing music.

Rob Ferrone

Sorry, yeah. And I hear people say, look, what's the playbook for this? As we talked about earlier on, tribal knowledge is an automotive startup, is an aerospace startup. What do we need to know? And everyone's got a slightly different perspective because they've got different backgrounds, et cetera. But to actually have not even necessarily the playbook specific to the industry, but like, what do you need in order to work together? What's the and if you if you don't even have to think about that because of what normally happens is you have teams coming in and creating the bureaucracy in the processes. But if you had this kind of capability, something that you is bringing to the party, you can actually get people working together in synchronized ways, very low friction without the overheads. Yeah, so I think, you know, it's going to it's going to lift startups and then it's going to, you know, make ⁓ that larger companies more agile and faster.

Michael Finocchiaro

But that tendency to break things first, to fail fast, that's sort of the whole thing in the new industry, is that when I thought of tribal arms, I was thinking more of traditional manufacturing where it's a bit slower and it's not, we don't want to break first, we want to fail later because we need to get the design done. It seems like there's big change or a big difference between. traditional manufacturing aerospace or automotive or industrial equipment and what you were describing, Ye, in terms of, you know, let's fail first and fail fast, fail first, and then do the next rev.

Ye Wang (CEO Of EverCurrent)

I don't think hardware can fail first and fail fast. You're really going to fail. I think it's really different from software. think a lot of ⁓ startups, I mean the Bay Area, so I meet a lot of actually robotic startups and so on. I think people are also having tribal knowledge, right? And people are relying on getting people with a lot of deep knowledge to come to the team to be like, how can we ship this thing with lower risk? How can we actually go to market? I think the difference Coming back to Agile a little bit that you guys mentioned, I think the goal is not to say you've this two weeks sprint, and you don't have gated process anymore. I don't think that's how successful companies I've seen running Agile do. I think they still have gated process, but they also have Agile. The thing is you always want to do the testing first. You want to understand the market needs you.

Michael Finocchiaro

Ahem.

Ye Wang (CEO Of EverCurrent)

Right. You don't want to this very deep, very long process. And in the end, right. Like the market has changed, right. there's not enough demand. I think, I think agile is good in the way that you can put something, maybe there's more frequent integration point, testing point, and so on that you can go to market. I think, I think that's a goal of startup. And I think large companies also need that. I think a lot of large companies are looking at how can they do you know, more innovative products that maybe have more software integrated into it. think traditional, lot of the manufacturers might be more on the mechanical electrical side, but then a lot of the margin of hardware comes from, you know, how, kind of software experience, subscription experience you can bring to your product. Your hardware might be collecting so much data and that's very valuable for your customers, but you just traditionally have no way to use those data to help your customers' business. Then they are looking at new ways also. And you know, software traditionally always kind of have been a long time running in Agile. Then people have this issue of, software is running this kind of process. How is running a very traditional process? How do we get people to work together? And how can requirement be passing? And you might not wait for like 100%, like all the requirement locked for you to move to the next stage. Then what do you do?

Michael Finocchiaro

But is this?

Ye Wang (CEO Of EverCurrent)

Anyway, going back to you, Michael.

Michael Finocchiaro

Is there a fundamental difference between when you're in this very fast paced, 21st century Bay Area hardware startup thing ⁓ to the fact that in the more traditional manufacturing arena, you've got an aging workforce that's gonna retire and that tribal knowledge is gonna walk out the door with them, right? That's a.

Ye Wang (CEO Of EverCurrent)

Thank

Michael Finocchiaro

maybe less of a concern in these fast moving hardware companies, but a really big concern with a lot of companies you've worked with Rob, right?

Rob Ferrone

Yeah, no, it is. But I think, you know, so there's multiple elements to this. There's, there's knowing how to work together and, know, having the tribal knowledge associated with that industry. If you're a startup, know, you might be working on something like the EVTOL, you know, craft that people are creating, you know, there isn't much tribal knowledge there, or even knowledge about that, that product and that industry. So they're kind of making it up as they go along. ⁓ But I think, you know, to kind of come back to the logic around what Yi's been working on, and Yi correct me if I'm wrong or not, but it's ultimately, ⁓ I think it's closer linked to lean, you like, but lean operations and, you know, really, ⁓

Ye Wang (CEO Of EverCurrent)

you

Rob Ferrone

serving humans and the people that are making decisions, the people that, that, um, you know, are manually tracking information or like, you know, you go into a meeting, you take, you take notes, and then you're relying on the person who's taken the notes to do them the followup. You know, for me, that's all part of the PLM because you're driving the product forward. And I think what you're proposing is that there's smarter ways and better ways of doing this and, you know, act like recording information, getting information to people, taking information out people and. and helping people to orchestrate the work that needs to be done. Is that right?

Ye Wang (CEO Of EverCurrent)

Exactly. Exactly. I can give a concrete example. Let's say you're robotics team, right? You're doing some kind of warehouse robots, right? Have a chassis, right? And then you went to a meeting and then you heard from, because like the product you heard from customer that, oh, you know, like the payload needs to be increased for this customer, right? A very typical thing that coming, but you're about to go into dbt, you know, and that requirement just dropped. So everything needs to move fast.

Rob Ferrone

Mm-hmm.

Ye Wang (CEO Of EverCurrent)

for you to adjust a bunch of things. And then, of course, like, chassis team get it first, because you deal with the payload directly. The chassis team got it first, had made a couple of changes, and they forgot to tell the battery team, maybe, and some other teams that will be directly affected, or testing teams that's buying equipment right now. And maybe your chassis is bigger, like, the equipment needs to be adjusted. In that short amount of time, A bunch of things can be missed, and that's kind of the gap we say, always kind of existing the, but who's gonna be responsible for catching up with those gaps? So with Evercurrent, what happens is like say, this meeting happened, you have a meeting recording. Like it automatically tells the product manager. Show up, like say, this meeting has happened, your requirements seem to not match what you discussed in the meeting. surfaced it. The product manager looked at it, oh, accept. Directly go to the AI pulled up on the right. We're meeting those cheat sheets. We basically provided you with the cheat sheets. And then your requirements on the left. You can just copy things in. And then when you make that change, the AI also said, oh, the requirement has changed. Notifying the engineering manager and the product manager. does this seems like a payload issue? Do you want to let your chassis team know? Right? And then you just accept. And then again, it just directly sends to your team. And then when the team made the change, right? And say, ⁓ your testing team actually preparing things and your testing team probably have a lot of things at a time before DVT to deal with all this requests coming from the supplier team, from your engineering team, last minute changes. all come together for the testing plan. Together. It's all bundled together. So the testing team just like, I can have the peace of mind. They don't need to ping all these different people. there anything changing? I overheard the session. But have the peace of mind. I sit here. I see a task list that all the people from product, supply, from engineering sent to me. And I also have the contacts attached. to all these different requests. And AI prepares some cheat sheets for me. And I can also include them in the same conversation until I close this issue. It doesn't really change fundamentally how you make decisions with the right information. I think the thing we do is we just want you to get those information easier, faster. Your things do change quickly in an organization. And that's when. Gap happens and big misses can happen as well.

Rob Ferrone

Yeah, and that's, that's ultimately real time configuration management. And, you know, you have, yeah, you have people these days in organizations who are the person in the room who says, on a minute, you're doing something here is, is luck, you know, you is lucky that I'm in this meeting, because I know the person that might be affected by this, and I'll go and talk to them, or go to their desk and what have you. ⁓ You know, and this is a smarter way of you know, doing that. So it's like always on, always listening, always plugged in, always connected. that's what. ⁓

Ye Wang (CEO Of EverCurrent)

And I think in the end, it comes back to the tribal knowledge of the whole team. I think people learn by doing. I actually believe people learn by doing. It's like if I'm a new engineer coming to the team, I might not know that. if AI told me, oh, this is why it's related, I learned that. I did that once, and I learned that. And it's automatic. It doesn't take the other senior engineer's time for the junior engineer to be onboarded to.

Rob Ferrone

Mm-hmm. Yeah.

Ye Wang (CEO Of EverCurrent)

how things are to be done. Where are the risks? How to avoid risks, introducing bigger risks into the program. I really believe people learn by doing, and then that's a really good way to just get the whole team understanding of the consequences of change much faster.

Michael Finocchiaro

There's actually a question from Elon Magyar in the chat. What is your take regarding the relationship between intellectual property and tribal knowledge? ⁓ How do these two interact within a business context? Are tribal knowledge and IP complementary or conflicting concerns? While I do recognize the benefits of tribal knowledge for competitive advantage, I also see potential concerns regarding the protection and sharing of IP related to internal processes. Who wants to aim at that one?

Rob Ferrone

Yeah, you're nodding. I think you've got an answer.

Ye Wang (CEO Of EverCurrent)

Yeah. I would say I can tell you a little bit. I actually don't quite understand the question directly, like which angle of concern. But I can tell you a little bit like how we approach IP and how we do that. And I would love to have the person maybe clarifying the question a little bit, maybe putting a scenario to help me understand a little bit more. ⁓ We are very, very careful about IP.

Michael Finocchiaro

Okay.

Ye Wang (CEO Of EverCurrent)

⁓ And the way to do that is ⁓ basically giving a team full control of what other things that should be flowing around and what other things that you should not. Things like a program. I would say a lot of times it's a lot of beneficial for everyone to understand the program. People are really very open to share the full program. Product requirements. I think people want everyone mostly.

Michael Finocchiaro

you

Ye Wang (CEO Of EverCurrent)

to understand what are the product requirements you are developing towards. People will track requirements. And we track the design changes, because those are the things traditionally people can get to see it as well. ⁓ So I don't know what it is. But then there were definitely secret sources, and you don't need to share. And you shouldn't share if you don't want more people to see it. ⁓ The other part of this is where ⁓ For aerospace customers, we are ⁓ basically ITAR compliant. So we're totally in the GovCloud. So we deal with the data with a lot of care. And for normal clients, more commercial clients, we are SOC 2. ⁓ Those are just the first things we're building day one. Data security is the most important things.

Rob Ferrone

Thank

Ye Wang (CEO Of EverCurrent)

But I'm kind of curious. I actually didn't quite understand the question as much.

Rob Ferrone

I if I just thinking out loud here, but so the you've got product IP, I ⁓ Yeah, which is. ⁓ Well, you mentioned secret sauce, I think, you know, I've always said that companies cook with the same water. You know, so a lot of people say, you know, ask me for benchmarking and saying, what does what does this other company do? And how do they do PLM, etc. But ultimately, it's, you know, unique to the company. And, and, and, but the secret sauce, then is the, you know, the softer stuff like the culture, the, the leadership, the, you know, the tribal knowledge that we've spoken about, etc. So that's an interesting one, because if you if you are capturing that, and that way of working and, ⁓ you know, the, ⁓ let's say the AI brain of that organization, I think that is a big differentiator. But as you say, you then you know, that information secure, then you should be okay.

Ye Wang (CEO Of EverCurrent)

Yeah, I think the other thing I think like, you know, a lot of say what brings a company success is innovation, you in the end is innovation is your speed of go to market with lower risks. Right. ⁓ So, so yes, you can hide certain things, but as long as it doesn't hurt your innovation. Right. But a lot of times is you want your team to feel more ownership as well. I think some of the big organizations, like I observe that the individual engineer might not even know what the thing he's doing contributing to the bigger milestone. There's a bit of a disconnection between what the engineer is working on to the company goal and so on. And that's the kind culture shift I really hope this new era of AI tools can help bring. People feel more ownership of the work they are doing. understand. And people, the whole organization, are working on the top priorities. People have shared understanding of the priority and so on. That just makes everyone feel more engaged and go forward faster.

Rob Ferrone

I think the point that I was struggling to articulate earlier on about IP is if you assume that the ⁓ IP relates to the product itself and the solution, whether it's a software or the hardware, I think the thing that needs real facilitation is the way that people work together, the way that people collaborate on that solution and that product. And therefore, the, let's say the actual you know, products and software, et cetera itself is less on the critical path in terms of the conversation we're having right now. And it's more about the how the way that people work together and the, the, ⁓ the smarter brain about, you know, not, not forgetting stuff, not being misaligned. ⁓ and so, but as I said before, I think that is part of the secret sauce. If you've got, you you look at race teams, look at the formula one, ⁓ pit stop teams, how, ⁓ practice they are and how fast they are because of, know, all of this stuff. the way that they work, etc. And that is secret sauce. And ultimately, yeah, but if it's secure, we're good.

Ye Wang (CEO Of EverCurrent)

The other thing I want to quickly add is ⁓ IP protection is extremely important for hardware teams to create a barrier, a competitive advantage for you. And oftentimes, filing that takes a lot of materials. ⁓ So one way people look at us is we can actually track a lot of the different things and ideas. This is for traceability, both for compliance and I think can potentially be IP.

Michael Finocchiaro

⁓ Elon re-specified his question in the chat. He said the risk of specific internal company processes becoming public. instance, a company utilizing a unique configure to order process would not want this proprietary information to be accessible to competitors because it's tribal knowledge. It goes to the way your product gathers data into the AI model and then it gets shared with the tribe.

Ye Wang (CEO Of EverCurrent)

⁓ yeah, yeah. no, no, it does not. does not. We never share data between organizations. So it is basically.

Michael Finocchiaro

So the tribe is defined as within the four walls of the company, not the tribe as in all aerospace engineers working on rockets, right?

Ye Wang (CEO Of EverCurrent)

Yeah, yeah, yeah, yeah. Yeah, yeah. for more, say, aerospace defense-related customers, use their own AI models. Some are hosting their own AI models. We can plug into whatever infrastructure you want. And then even if you're using, say, bedrock models on AWS, it just stays there. It does not go to another. We never cross-train between customers. That would be disastrous.

Michael Finocchiaro

But it makes me wonder, how do you separate the stuff that needs to go into the system of record from the stuff that's just tribal? How do you differentiate between that information that's very important for the bomb in order to actually build the thing and stuff that's more wishy washy, a little bit more ⁓ touchy feely on how processes work, guess. How do you sort those things out? Rob, do you have some ideas on that?

Ye Wang (CEO Of EverCurrent)

Thank you. Mm-hmm. In some way, we

Rob Ferrone

Wait,

Ye Wang (CEO Of EverCurrent)

do not. But I'm kind of curious what Rob saw.

Rob Ferrone

Now, let me predict your answer here. You tell me if I get it right or not. I assume that things like ERP, PLM, cetera, ⁓ PLM as a system and all the other systems record will continue to exist and that information will exist in those places. And then what you're doing is they're more facilitating the... ⁓

Michael Finocchiaro

Do a thumbs up or thumbs down, right? ⁓

Ye Wang (CEO Of EverCurrent)

Yes.

Rob Ferrone

consolidating that information and checking it and flagging any disconnects and flagging things to people and saying, hey, have a look at this because what you're showing in the e-bomb is different to what you're doing in the e-bomb and that kind of thing.

Michael Finocchiaro

you

Ye Wang (CEO Of EverCurrent)

Exactly. Thumbs up. Yeah, I have prediction. feel old systems are still going to be there. But people need a modern experience. ⁓ If you want to increase the capacity of the team, you don't want people to. And younger people, think, especially, would not want those kind of jobs, just checking the gaps between systems.

Rob Ferrone

Yeah. Mmm.

Ye Wang (CEO Of EverCurrent)

You know, like in your private life, you're using AI that like everything's going so fast and they don't want in the work life to be doing jobs. That's it's too prescriptive. Right. So really what we do is we just, we don't want to create an interruption to your operation. We sit on top of your existing system, providing a more modern experience. the experience you provide is like, we understand all the difference, like all the changes you're making. And we just try to orchestrate the change, making sure the dependency, the people who have dependency on the change also get to know it. And then when people get that change coming as a request, and we make sure we prepare some cheat sheets and getting you also the context around that change and translating that. I think people don't realize it's in Harvard. There's a lot of translation happening. Your surprise team doesn't always understand all the things your engineering team is talking about. So doing that translation, preparing a cheat sheet, getting things just a little bit smoother. People can focus on work. AI brings information in.

Rob Ferrone

Yeah, and this is, this is, you know, I posted the other day about the total cost of doing PLM. And when people think about PLM, they often talk about, you know, PLM systems, as I said. And, you know, if you compare PLM systems, okay, maybe you can save a tiny bit of money if you switch from one system to another, you know, but the thing about if you could, if you can make your workforce, you know, 30 % more efficient, or if you could do, you know, 50 % more with the team that you have, you know, that's huge. We're talking big money there. So, you know, this is this is the, you know, the part of PLM that people aren't really thinking about, but they actually should be.

Ye Wang (CEO Of EverCurrent)

Yeah, I think this already happening in software, like with AI, right? ⁓ and I was reading the same reports about like a startups. ⁓ so people who are, you know, say graduating from seed, going to series a the size of the company, like the busy people getting the same amount of investment, but the size of software company. It's like half of nearly happened. Right. It's really happening in my day to day work. Right, just how much an engineer can bring to the table, how fast we can put something to address our customers' needs fundamentally shifted. I think hardware will welcome a moment like that as well. But you need to get your knowledge together first to welcome a moment like that. Yeah.

Michael Finocchiaro

It's interesting because I've talked, I talked yesterday with the SysGit, so they're doing ⁓ requirements management, but basically using Git because they think that, they, ⁓ Steve Massey, the founder recognized that ⁓ software is much more agile, much faster to adapt than these old kludgy, clunky requirements management systems, which reminded me also of how, and now I'm gonna lose my train of thought. ⁓

Ye Wang (CEO Of EverCurrent)

Yeah.

Michael Finocchiaro

Just I've seen a lot of other ⁓ startups doing things that were really, really innovative around that of just trying to accelerate ⁓ and imitate some of the success in software so that hardware companies could move faster. I think in the pre-show also you were mentioning something about quality. You wanted to talk about quality and travel knowledge.

Ye Wang (CEO Of EverCurrent)

Mm-hmm. Yeah. Yeah, definitely. I think in any hardware company, they're always trying to balance ⁓ three things. You're balancing quality, you're balancing schedule and cost. I think traditionally, some people say that you can only pick two out of three to focus on just because of how much resource you have and so on. I think there's a big story to tell about quality around how AI can help. having traceability, or actually implementing the V model in a lot of different places. It's actually really useful for the organization. But nobody has that resource to do those kind of validation and to really track things in meaningful ways. Like, it requires a whole team to have a total different level of documentation. Medical device companies spend a lot of time doing that. because of the requirements. But can that be beneficial for other companies? Probably in some degree, yes. But people do not want to have that cost, spend that cost of totally changing organizing to do documentation. I think the way I look at opportunities, who we power. We're really well on power. First of all, of course, engineering managers to unlock the capacity of the engineering team. We want to power TPMs. who are just always doing this integration. Our integration teams doing this integration is a very difficult job. And then we also want to really power system engineers. I think even with one person with systems like ours, what we're hoping to see is it can really make a lot of impact on the full organization by reducing the risks of the products you're bringing to the market.

Rob Ferrone

To preempt a question from Michael, what does this mean for young people getting into the industry? this taking away jobs from young people? Because they would classically come in and they would have to learn the ropes, have the mentors training them. They would also do the junior roles often just to get up to speed on the business. And is this actually advantageous to them in the sense that they could come in and have a faster impact on the business? Or is it taking the work away from them in the sense that they won't have the kind of graduate and entry level jobs? question to you, Yee. That's the question Michael would ask.

Michael Finocchiaro

Yeah. My next question. So, said he preempted it. Thank

Ye Wang (CEO Of EverCurrent)

I was waiting for Michael to... Yeah.

Michael Finocchiaro

you.

Ye Wang (CEO Of EverCurrent)

I can tell from like my personal experience building my team, you know, I have very senior engineers, but I also have very junior engineers. And the impact that junior engineers are making, you know, like new grads from college are making. I will say are much more than what I had back in the day when I first graduated. I think people bring a lot of new thoughts. I think new generation bring so much new thoughts. They're very much more native to AI. ⁓ They are very good at using a lot of different tools. And then I think people who are good engineers get more opportunity. to work on, focus on the engineering problem. So I don't think, like I think people who are very good with AI to kind of multiply their powers by several and maybe two today, or maybe more, even more tomorrow, who knows? I think we'll always have opportunity. And I think also what you see, what's very interesting in the market is also there's more hardware startups. And not all these founders had, say, 10 years of hardware development experience. Some hardware startup founders, they might came from totally non-hardware background. And I think that's going to be a lot more because people have their personal experience and knowing this problem needs to be solved. Let's say like last mile delivery of robotics. People who understand that problem so admittedly might want to work on something to change their world. And I think AI will empower so many people. I think more people can come into hardware and bring innovation to the physical world. And that's what we're excited about.

Michael Finocchiaro

I've ⁓ interviewed quite a few startups in the last few months and I always ask that question. And a lot of people have, ⁓ of the founders have said, well, ⁓ the younger generation needs to focus on the fundamentals because, know, no AI ⁓ is going to tell, is necessarily going to be able to, to, ⁓ to interpret the result and tell you whether that was good engineering or not. Right. And that's really only from doing the math and having gone through it that you can look at something and immediately say that that's a hallucination. There's no way it works.

Ye Wang (CEO Of EverCurrent)

Yeah. I would disagree with that a little bit. Yes, definitely fundamentals is important. You need to have very good judgment around your thing. But I feel AI will get smarter. It will also understand the fundamentals, like say in a couple of years. Very well. I work out all of this research. A lot of the research we're doing are more like cutting edge. Not in the real world yet, but But it's pushing the boundary. So I know things like, fundamental, like, geodesign probably can be solved with AI very quickly. ⁓ I would say engineering training is something that's meaningful. Engineering training is not just a training about fundamentals. You understand this equation. You can derive this equation. You remember this equation. But it's more around, like, this is a problem. How do we solve this? How do you prioritize it? How do you the trade-offs? I trade-offs in hardware are so important. ⁓ traditionally, think as an engineer just graduating, you might focus only on electrical, or you might only focus on mechanical. And how do you have big impact in a robotics company that has mechanical material, like all the different functionality? think the speed of your learning is slow. But I think today, people who have that curiosity, who have that drive, can learn so many more about all the disciplines they are working together at a totally different speed from before. I'm really optimistic. I feel like people who want to learn, who want to have an impact, who want to do good engineering just have so many more tools today to support them. then you say, traditionally also like writing test plans. It takes engineers some training to write good testing plans, because you might not understand what can be tested. Or like, say, certificate, getting certificate here and there. There's a lot of implication of what you can touch, what you cannot touch, what takes long and so on. Who's going to those documents? Who loved reading those documents in the past? But now AI can bring a lot of those knowledge in much faster. So some of the things we do is also adding points in your plan of, say, gap analysis, deep research, checking if you're required. In the big environment, anything has changed. Because it's not the best use of people's time. We want to bring more information to people so they can learn at a different pace.

Rob Ferrone

I'll give you a really concrete example of an optimistic example. And the company I founded and grew, we used to take junior people, sometimes, you people with biology degrees, and we would bring them on board because what we found is we couldn't we couldn't find product data managers, because everyone was either, you know, specific to a system, but actually, agnostic product data managers, they didn't exist. So we had to create them. And you'd find very, very bright graduates from good universities who thought in a special way, but also were great with people and communication. And you brought them on board and they went through a four week bootcamp. They had support from their peers. They had continuous learning and development. And these people were going straight onto project and facing off to very senior industry types. But we found that actually they became the kind of experts in the room on the best way to execute a product data management strategy, even though they, you know, they had less than a year under their belt. But it was this, ⁓ you know, the right information at the right time to these people, you know, combined with the DNA of the way that they thought and the way that they approach things. And actually, you know, that was the superpower. That was the reason why, you know, these people were so cherished and everyone wanted them in the organizations, you know, they were running so. So I do believe that there should be some optimism around, you know, young people and if they have the right toolkits and the right support as they come into these industries, they can actually be game changers for the companies that hire them. So you don't have to have, you know, a CV with 20 years of experience. That might even be you know, a detraction.

Michael Finocchiaro

That's because they're thinking somewhat outside the box and they're not constrained by 20 years of doing something the same way.

Rob Ferrone

And they're really focused on that specific topic, know, that niche, they didn't, let's say they didn't care about the engineering itself, they cared about how to support the engineers and how to support the product information flow to get the product out the door on time at cost and at the right level of quality. So that was their focus. And that was their niche. That was the thing they excelled at. And again, it was that idea of deconstructing the task. If you deconstruct the task and give it to the the person or the technology that can do it best, you'll get far better performance than if you, you know, have to kind of force people to do that work. You know, engineers are being forced to do these kinds of tasks, know, meetings, like information searching, we said, Excel spreadsheets, systems, you know, with, you know, as part of their daily tasks and they're just the wrong people for it. They should be engineering. They should be focused on, you know, what they do best.

Michael Finocchiaro

Mm.

Ye Wang (CEO Of EverCurrent)

Yeah. I think the other thing is people sometimes are limited by what they think the speed of learning could be. So a lot of times it's like you put younger people in an environment that you tell them there's no limit. I think you can do this in one day. They might really turn something in one day, maybe not the perfect one, but they did that in one day. And then their expectation of speed is totally different. I work with amazing engineers at Onshape. have worked at SolidWorks, have long career. They're the most curious people. The other people don't feel like there's a bond to how fast you can do things, how well you can do things. And I think that's, with AI, we just have tools. I think it's exciting for people like that because it's like, wow, there's really boundless and AI is only getting smarter. like my ability of doing things also get a little bit boundless, There are always going to be people who jump out of the traditional spectrum of doing things, right? You'll always have a place for innovation.

Michael Finocchiaro

So we're getting ⁓ close to the time. I'm wondering, what do you guys think about the future of tribal knowledge and AI as we move into 2026 after experiencing the LLM explosion of 2022-23 and the MCP explosion of 2025? I mean, what's next? how is that going to impact this whole tribal knowledge topic?

Ye Wang (CEO Of EverCurrent)

you think it's really going to spread in 2020. It's already spreading. think maybe hardware is a little bit late compared to, say, security. You see, there's so many knowledge systems. If you talk about knowledge systems, it's actually a well-understood thing. I think it's coming a little bit late to hardware. And then, I want to add a little bit of things. So in January, we're going to be at CES. So if anyone listening to this want to come to CES, say hi or reach out to me on LinkedIn. Next year, we're also going to spend time in Boston, in LA, in Vegas, in Seattle, in a different, and in San Diego. You wish you come to Paris. That might be a leisure trip. Yeah, I'm also going to be in, you know, like Texas, New York. So, you know, if you are interested in, like, you know, meeting us in person, let me know.

Michael Finocchiaro

Nice. You're to come to Paris? You're going to come to Paris? We're ahead of MSA. That's closer to Rob. Thanks. Ravi, any closing comments?

Rob Ferrone

⁓ I, I guess it's, I think the people that are going to have a significant shift on, ⁓ business performance are going to be the people that really think about the way that work gets done. And, that is, is a big unexplored area. And I think, ⁓ when you see the big breakthroughs in terms of what companies achieve, it's going to be those companies that have thought about this and deconstructed it. and change the way that work is being done. yeah, that's the thing I'm really excited about. You know, I've been living it for 25 years. You know, we've been the breakthrough in terms of the human part of that. But like, if we if the work we've been doing can be replaced even ⁓ better by the AI and the types of things that used developing then that is really exciting.

Michael Finocchiaro

Well, it's a real pleasure having you go put you guys on. It was nice talking to you again, yay. And always a pleasure, Rob. ⁓ Next week I have another webcast, this time with SisGit and ⁓ Epsilon 3. So was more about requirements than about manufacturing. And once again, thank you very much to Yi Wang and to Rob Ferroni. And ⁓ you guys wanna say goodbye?

Ye Wang (CEO Of EverCurrent)

Happy holidays everyone!

Michael Finocchiaro

Yes, it's a Christmas edition. And we'll see you on the next podcast. Yeah, I'm trying. I'm on the mend. I'm on the mend. Okay, thank you everybody.

Rob Ferrone

Yeah, get well soon as well Michael. I hope you feel better before the new year.

Ye Wang (CEO Of EverCurrent)

I'm gonna get...