Sam Zheng Podcast Transcript
Sam Zheng joins host Brian Thomas on The Digital Executive Podcast.
Welcome to Coruzant Technologies, Home of The Digital Executive podcast.
Brian Thomas: Welcome to The Digital Executive. Today’s guest is Sam Zheng. Sam Zheng, CEO, and Co-founder of DeepHow spearheads, a rapidly evolving startup act by esteemed investors, DeepHow revolutionizes skilled workforce training with an innovative AI powered video-centric knowledge capturing and transfer platform.
Prior to DeepHow Sam dedicated over a decade to Siemens driving digital innovation across various industries. His noteworthy projects, such as the cloud digital inspection jacket, have significantly improved technical knowledge sharing, efficiency, and user experience, earning his team the prestigious Siemens Innovation Award.
Well, good afternoon, Sam. Welcome to the show. Thank you. Thank you, Brian. Absolutely love doing this, Sam. I appreciate you making the time hailing out of the East Coast there. And the time difference isn’t so big today for this podcast but love to do these things every single day. And Sam jumping into your 1st question after over a decade at Siemens driving digital innovation, what inspired you to co found DeepHow, and how did your experiences at Siemens influence your vision for DeepHow? Sam.
Sam Zheng: Yeah, absolutely. So, the Siemens experience actually really set the foundation for why we started DeepHow. Tell you a little bit of background, even before I worked at Siemens, I actually studied engineering psychology.
Always passionate about understanding people, helping people, seeing technology is playing such a big role. Significant role impacting every aspect of our life and work. That’s I want to study the intersection between technology and people and how to leverage technology to help people. And after I did my PhD at University of Illinois at Urbana Champaign, and many years ago, then I joined Siemens Corporate Research and leading many digital innovation projects and working very closely with our partners in manufacturing, in technical service, industrial service.
What we have realized is majority of these innovation project, digital innovation and digital transformation and so called industrial 4. 0 projects, they were focusing on introducing more automation and robotics to the shop floor, to the industrial environment. And yet we often hear from the managers, operational manager and plan managers saying, Hey, you know, robots and automation can actually cannot solve all the problems.
We love them, but we, we are actually desperate to have more skilled workers. And that actually, it’s really because many of our experience, you know, experts, like baby boomers. And after spending years on the shop floor and in the field and getting all the knowledge and they’re retiring and their knowledge has not been effectively captured and transferred to the younger generation of workers.
And speaking of the younger workers, they grow up learning very, very differently. They are no longer interested in reading the lengthy manuals, instructions, SOP documents. And in fact, if they want to learn anything, they prefer to go online to watch videos because it’s just more engaging and more interactive.
Go to YouTube, your TikTok. So that way it made us think, well, how could we use technology in this case, but give you your AI technology to capture experts knowledge and digitize them and deliver in a way that’s engaging to the younger generation of workers. Can we develop technology to retrain people quicker and more effectively, more efficiently?
And that will lead us to create DeepHow. And DeepHow, that’s exactly we’re leveraging, you know, technology and particularly AI and video to capture, you know, expert knowledge and then deliver that to the other workers and make it much more engaging and easier to learn.
Brian Thomas: Really love that. Obviously, you have a knack for understanding human psychology, right?
And having that background, I think, really helped shed the light on a need here. And you’ve done a great job of doing that. So I appreciate the, the backstory on that, Sam. Sam, could you explain the AI technology behind DeepHow’s video centric knowledge capturing and transfer platform? And how does it specifically cater to the needs of the skilled workforce?
Sam Zheng: Absolutely. Absolutely. AI now is such a buzzword that gaming, I mean, the technology is advancing so rapidly, right? But in recent years, you know, powered by this large language model like ChatGPT. So, we started the company actually five and a half years ago, and we’re already leveraging all the state of the art, you know, AI technology.
In our particular case, it started with digitalization. Is we want to capture your experts, your knowledge. So, the best way at your most effective way, actually, is to use video. Because today, if you look at a lot of these, your expert’s knowledge is captured. You’re using these text based documents or instruction manuals or SOP, standard operational procedure.
So someone has to write it down and either the expert they have to do by themselves or someone else talk to the experts saying, Hey, how you’re doing, you know, certain procedure observe them, et cetera. Actually the, the more efficient effective way is, well, why don’t we just, we call expert, you know, doing things by everyone actually on the shop floor that everyone has cell phone.
And then there are also designated mobile devices I use. And we already see people doing that, right? Start doing more like video recording, you know, capturing expert, you know, doing work. And, but then previously it takes a lot of time trying to organize these kinds of content. And that’s where AI comes in.
So, we developed an AI system, right? Already building into our mobile app. We call it a workflow capture app. Literally as expert doing demonstration. And then someone can just record the expert or all the expert can do self-recording. And what AI does eventually turning this demonstration into a step-by-step instruction.
So that’s a breakdown to see what AI does here. Number one is now it’s becoming so obvious, right? So, I can transcribe your expert. You’re speaking actually different languages in a very noisy environment. We’ve developed this specifically for this manufacturing and industrial environment. You’ll have experience working there.
It is very noisy and sometimes can be very loud, right? Over 100 decibel and people need to wear all the hearing protection, but we develop technology and combine it with software hardware. Now we’re able to capture experts speaking in this noisy environment. I can understand and I can also understand.
You’re more than 50 languages, so that really makes it very easy for particular company if the expert you’re speaking one language. And then later, I will also translate that into over 50 other languages. So, we made the knowledge transfer very easy. And most importantly, one of the key thing we do here is, of course, AI does all the indexing.
We’re using natural language processing and also your computer vision or so called multi model because you want to your leverage, you know, all these different modalities and understand as expert by doing a, that, that’s a assembly and certain machine or fixing certain parts. It could be, that’s a 30 minute or longer procedure.
Our AI would come in. And trying to understand what are the key your steps, right? Extract this key information, and this is what we call your workflow segmentation and turning complex. Your demonstration, break it down and then into by size and your, your step-by-step instruction. And that really may make it much easier for people to learn, right?
Because just like, go to your YouTube and, and now it’s becoming popular. Now people start creating chapters. So, there is tons of research to show. So, we can break down those complex topics into, you know, these step by step or chapter by chapter. And then it’s a lot easier for people to consume.
And then also it’s a lot easier for people to search and then to go back to the only in the places that they need a little bit of refresher. So, in short, what our AI does is turn experts Your demonstration and into your step-by-step instruction, and we keep also human in the loop. And in fact, experts can review everything what AI does and if they are okay, and I can publish this.
And by the way, we also use AI to generate quizzes and so that the learner, right. So, after they watching the video and then they can also take the quiz and these quiz again, will reflect all the key points and key steps. And that our AI has already been extracted. So that’s really just in the nutshell, turning demonstration into very organized step by step instruction.
Brian Thomas: That’s amazing. Sam, I appreciate you obviously recognizing that video is the way to go for training, but then you infused AI into this that is actually able to overcome some of the noise complications that you have in recording in traditional manufacturing and really highlighting just how AI simply is getting it done much faster, more efficient for workers, especially the new ones that are need to learn the process.
So I appreciate that. And Sam, looking ahead, how do you see AI and machine learning evolving in the context of workforce training and what trends or innovations are you most excited about?
Sam Zheng: Yeah, absolutely. AI is going to play such a critical role in workforce training because in recent years, AI empowered by deep learning or deep neural network and has managed so rapidly, right?
The recent AI development, particularly, you know, like large language model, which power your chat GPT at the beginning shocked the whole world, right? So, well, AI, it’s now becoming really capable. So. And there’s also a lot of fear by people in general of people you’ll have saying, Oh, wow, is AI, it’s becoming so powerful.
Are they going to replace our jobs? Right. And then would that create, you know, all these problems. And for us is definitely AI is such a powerful technology. We need to use it wisely. In this case, we should definitely leverage it to help people to get better. And that’s why AI will play such a significant role in workforce training, and specifically what we think, why AI and particularly today, like generative AI can play such a role.
Yeah, I just mentioned, right, the, the AI can turn this unstructured demonstration into more structure, your way of delivering step by step instruction and demonstration to step by step instruction. That’s one part. And second is AI actually can create a lot of these more personalized content for any learners.
Like for instance, so after we capture all this information, of course, your people can just, you know, view it by the similar experience they have on YouTube or TikTok. So, our AI will provide also recommendation, but in this case, we’ll tailor to their career development needs, skills, development requirements, right?
So, we’ll use recommendations to really provide the right content for them, make it easy for them to access, but now they can also ask questions directly. Okay. Where’s some of the like, in this case, we actually wanted this AI to the data that we already created. So, this is so called retrieval augmented generation is RAG. Right?
So, so imagine now we already have a, a huge, that’s after capture all these experts’ knowledge, and then you have already the digital version of the expert available. Now from the learner’s perspective, they actually can interact with the system. I’m just asking any questions. And then AI will, of course, be number one, understand the question and later, you know, search all the related results.
So, what does RIG, Retrieval Augmented Generation, it does, instead of just presenting back, hey, these are the, you know, whatever, your 50, you know, related, you know, documents or videos and AI will actually synthesize all these content into more cohesive answers. So that would make actually learning a lot more interactive, a lot more personalized and a lot more, a lot more efficient and effective.
That’s another actually very important project where we have been working on is we call knowledge mapping. Just like today, when we travel, we rely so much on all these, you know, digital maps, like, for instance, Google Maps, right? And particularly, you go to a new place, right? At first, right? So, without these maps, it will be so, you’ll get lost.
It’ll be so overwhelming. So, our goal is now after we capture a lot of this knowledge and then manage AI to really make sense of all these information and then create and some kind of digital map allow anyone, right? No matter how you’re new or into this environment, they can leverage this map to navigate and without getting lost.
So that’s really something we’re super excited thinking AI can we help people right and better navigate in this complex environment in complex domain, but also be able to learn in a more personalized and more efficient, more effective way and leverage AI to really help people instead of replacing people or workers.
Brian Thomas: That’s awesome. Thank you for highlighting some of that. You’re right about the AI. AI can be available 24 7 doesn’t call in sick and be able to intelligently answer questions when students have questions. So that’s I think that’s simply amazing. And I appreciate you kind of sharing some of the innovations that are going to be coming about.
And Sam, last question of the day, if you could briefly share, how has your background in engineering, psychology and statistics shaped your leadership style at DePauw and what principles guide your approach to team management?
Sam Zheng: Yeah. So my background certainly has profoundly impacted how we approach this, right?
First of all, so we always emphasize AI, in this case, help people and we are taking this with a human centered approach. So human centered AI, that means the very powerful technology is how do we leverage it to help people understand, right, in this particular case of manufacturing your customer and then the workers on the shop floor.
So what are their specific needs and how we can leverage technology to help them. In fact, our mission is to really empower workers. by skilled workers to excel, right, by leveraging AI technology to capture the expert’s knowledge and then transfer that knowledge, you know, efficiently, effectively, and then to achieve operational excellence, right?
The mantra we have been using is, it’s called capture expertise and cultivate excellence. So that’s really the, the, the mission and that, that really inspire us to do DeepHow. And right. So, the background, what made DeepHow unique is we’re very human centered. And we’re, we’re a technology, we leverage all the technology, but in the end, the key thing is we want to get help people to learn more efficiently, more effectively.
We also want to help the organization to become more operational excellence. But the key thing here is really, you need to have both humans who should still be in the center, leverage the power of technology. In this case, particular AI technology.
Brian Thomas: Thank you again, Sam, for sharing your background in engineering psychology has obviously, we talked about this contributed to a broad need that will solve a lot of training issues and safety issues in the future in the manufacturing space, of course.
So do appreciate that. And Sam, it was such a pleasure having you on today. And I look forward to speaking with you real soon.
Sam Zheng: Thank you so much, Brian. It’s a great pleasure talking to you.
Brian Thomas: Bye for now.
Sam Zheng Podcast Transcript. Listen to the audio on the guest’s podcast page.