Go back

E258 Turning Semiconductor Data into Smarter Manufacturing with yieldHUB’s Gillian O’Donnell

from AI Ireland with Mark Kelly

14m 33s

E258 Turning Semiconductor Data into Smarter Manufacturing with yieldHUB’s Gillian O’Donnell

Yealtube is a leader in semiconductor manufacturing analytics, offering a data-driven platform focused on yield, quality, and reliability. Its Yealtube Live product deploys lightweight AI agents on high-cost test equipment—without disrupting production—to provide real-time visibility into manufacturing performance. This enables immediate corrective actions, improves first-pass yield, and maximizes existing test capacity, helping companies defer capital expenditures. The platform centralizes fragmented data from testing and manufacturing, eliminating silos and providing accurate, up-to-date insights. Advanced machine learning models analyze historical and real-time data to predict failures and recommend optimal actions. Yealtube is now expanding into an AI-native ecosystem (Yieldtop AI) that allows customers to develop their own AI models using real-world data. The solution not only enhances operational efficiency but also fosters human-AI collaboration, enabling workforce upskilling and better decision-making. By making data accessible and contextual, the platform supports smarter, more responsive manufacturing—critical in a high-demand industry where transparency and scalability are key.

Transcription

2484 Words, 13915 Characters

English
We specialize in semiconductor manufacturing and what we focus on is yield, quality and reliability. That saves engineers a lot of time, saves company time and it helps them with their yield management. And there's about, say, curative chips in your phone and about a thousand in your car. So all of those chips have to be tested before they go out into the field. And each one of those units goes through a tester that costs about a million dollars. We really need to empower the people as well. It's not just about replacing humans. Humans and AI can work very well together. Welcome to the AI Ireland podcast. I'm your host, Mark Kelly. delighted to say that we've got Jillian O'Donnell. Jillian is director of growth at the Year of Hope. And today's podcast is sponsored by Technology Ireland ICT skillness. If you're interested in finding out any more about education, digital and skills, check out their website. Jillian, thank you very much for joining us on the podcast today. Thank you so much for having me. I'm really excited. So we're going to talk about yield. Because they are finalists for this year's AI Awards. We're really, really excited about that. But before we do that, how did you get into the world of AI and data science and in your role, EO? Sure. So I studied data science and analytics at University College Quirk, which has now changed to data science and AI, but it's fundamentally the same course. And during my time there, I interned in various companies, software engineering, infrastructure engineering. And then I kind of realized that I wanted to go into data science. And during my final year project, I built multi-agent reinforcement learning for cyber security defense. And I realized during that, I kind of entered a business competition and I realized that I love the business side, the communication side of it. So when I started at Yealtube, I actually began as a data scientist and I worked on the product that's actually nominated Yealtube Live for agente AI innovation. And I worked in that. And then after about a year and a half, then I started moving into business development. So now I work across partnerships, sales, helping with technical marketing. And yeah, I love meeting all the companies and customers and stuff. So that's kind of what I do now. Yeah, so-- I really like that you've gone from that such a good knowledge base in the world data science AI under that data engineering. The nuance of the problems that you're kind of working on. But then working on the customers to say, actually, getting more solution-oriented, talking to them, listening, what their pain points are, and then coming over that kind of solution. I think it's a really, really good trend of what's actually going to start to happen more. Because those soft skills are really required, particularly in the world data science, where I can do so much more to help data scientists to do their work. So for people who don't know, what do Yealtube do? So Yealtube, we're an advanced big data platform. We specialize in semiconductor manufacturing. And what we focus on is yield, quality, and reliability. So we essentially take all the data across semiconductor back end test, back end manufacturing. And then we centralize that all into our platform. We cleanse it. We automate the analysis and reporting. So it saves engineers a lot of time, saves company time, and it helps them with their yield management. So we target companies, I guess, that design, manufacture, and test chips. So we've a wide array of customers across the world. And we have a global team that supports these customers. When we think about that for your average customer, having not having transparency of what's happening until the test is over, how can Yealtube help with that? Yealtube Live is a breakthrough product in this area. At the moment, capacity for these test factories is a big problem. Because there's huge demand for chips and they're trying to meet their customer demand. And actually, well-funded chip startups at the moment are finding it hard to get space in these test factories. So essentially, what Yealtube Live does is that it exposes issues during production to the test floor, during production, so that they can make the corrective actions that will speed up their manufacturing, speed up their test, and maximize their capacity. So we've actually had customers that have been able to defer capital expenditure on new equipment because of this. - Amazing. So you're giving more transparency. You're giving that kind of real-time information, which is so important. So then you can make kind of better decisions as well. For a customer that's kind of using us, how does the technology act as an enabler? What would enable them to go? Okay, I'm going to use this, and I'm going to be better off. What's the magic sauce that's happening there? So Yealtube Live is a very unique type of software and it's a real breakthrough in the industry. So it's a lightweight agent that we put on, say, every tester, and every one of those testers is about a million dollars between 500K and 2 million. So it's really expensive equipment, and you don't really know that much about it, but with this agent on the tester, it's very lightweight, so it doesn't actually impact production. In the industry, you really, we really stop ourselves from ever, you can't impact test time. So this is a very seamless software that doesn't require any updates to the test program, it doesn't require a new hardware, and essentially it streams clean real-time data during production to a, I guess, on the server side into a recommendation system that processes all this data with historical context, product history, contextual production environment, and other failure histories and economics, as well, of how much it costs to test this product, the unit cost, the retest cost, all of that, and then we provide recommendations in real time to the people on the test floor who are beside these machines that can perform the corrective action to maximize their good parts per hour yields and their utilization. So it's a breakthrough in that sense because it can help maximize their first past yields, which is a critical KPI for a lot of these test factories to meet customer requirements, and then also the capacity, because capacity is an issue, it maximizes the equipment that they already have so that they can defer their CAPEX. So yeah. - So you're giving them access to a dashboard that they can see in real time, what's happening. It gives them the opportunity rather than not have additional sun cost because you can write source pivot or change things in real time to actually ensure that you're going in the right direction. And for us, the analysis is probably going to be helpful for either now casting, but also forecasting, because you've got that access to that data. And then you're cleaning it, right? You're kind of making it accessible. Accessible, it's one source to the truth. So that in itself is a big benefit for a company. - Yeah, it does prediction. So it will actually predict if the equipment is about to misbehave or if it's going to be low yielding, that you can actually just fix any setup issues. So that's very helpful for them. Also, the fact that it cleans data, it provides true real time data because these factory systems often are too delayed or they might have some manual updates that you can't rely on that for trustworthy AI. So we can actually provide this real time data as well as you can have the recommendation systems that you can actually build really reliable AI models. So our customers actually can do that as well. - It blows my mind that only 5% of our listeners actually subscribe to the podcast. We've got one favorite ask, please can you subscribe to the podcast, if you like our content. You subscribe, we will make a huge difference to the guests that we can get on the show and also be a voice for AI applied across Ireland for doing it the right way. Now back to the show. - One of the things I've kind of learned from the semi-conductor industry and the whole world of manufacturing is there isn't enough industrial grade data to do enough of these kind of AI projects in our models because it hasn't been necessarily collected. So having that data is so important for future AI projects for people to work on. - Oh yeah, so we join our customers from their first silicon to write through to high volume production. And now we're actually on this journey of digital transformation with our customers and we're on their AI road map. So we are actually building, we have a Yieldtop AI which is our AI native platform. So we have Yieldtop which is very advanced big data analytics with machine learning, et cetera. But then now we have this AI native one that has AI assisted engineering. So we actually are on that AI journey with our customers and with Yieldtop Live they can use the data that we provide to build their own AI models too. - And that can potentially transform the whole business, the whole way they're actually doing business which is so exciting for them to go ahead. - Yeah, a lot of these companies want to build the smart factory they want to really lean into smart manufacturing and AI can be slightly ambiguous to them at the moment. You know, some companies really are all about AI and then others don't they just want to make sure that what they have is very trustworthy because we're going to vary and really regulated industry. Yeah so yeah I think with Yelto Blive it's because I guess we have the expertise in semiconductor test and that we've been able to build this really specialized platform and yeah and like even our CEO gave a talk a few weeks ago about AI and he says you know we really need to empower the people as well it's not just about replacing humans you know humans and AI can work very well together you know you can upskill your workforce while upskilling the AI and you know if you have really high turnover that's not that great because you're losing that knowledge. Yeah an idea is so important because you're teaching the AI skill but also the AI is teaching you a skill so the human is actually growing and using AI to learn how to learn faster. What I love about what you're doing is you're enabling the people to have better access to data be a more productive and more efficient on the kind of blue euro and in the green euro you're now offering them new opportunities for growth you products or services that they could bring to market and in that smart factory where things are real time it's all kind of data led and you can actually have one source to share it at any one time it's an opportunity to 5x10x what the company might do in the future because it's grounded in the truth of what's actually happening what's happening now for the vast majority of companies can't get clean accurate data it's not accessible it's not governs a lot of people are still using paper and not just as a big bot and that for scaling. Yeah a lot of companies still now have data silos and they'll have you know some people they'll have different engineers using their own maybe their own custom scripts or spreadsheets and things are kind of they're in silos and we break we kind of the benefit of having a platform like ours is we centralize all that so if someone leaves the company you still have all of the information but another thing about yield hope is that we have a knowledge base so we have a customer that we've had for over 16 years and we have every so it's so though that customer can actually add comments onto the data images voice notes videos pdf context and we have this layer that is just incredible for now that we're building our AI native platform and we have this context layer for these for our customers and you know it's really that's really important because if there's an issue with a product that's in the field say a chip and you know there was an issue with an airbag or something in a car and that chip was sent back and they were like what's going on you actually have that information already on the yield hope platform so that's so helpful having that ground a trusted accessible source so that you can always kind of leverages as well and thank you very much for coming in and joining us in the podcast and wishing the best luck and this year's AI Awards okay thank you and AI Island we're helping over a million people become literate using artificial intelligence and we do this in a variety different way the easiest way is our AI electricity program where we do a 60 to 90-minute program via virtual zoom or Microsoft Teams where we teach the rules of the role of AI from basic business concepts to ethical and resourceful decisions and how you have to use AI every day this is our literacy foundation the second way we use AI training is within our workshops these are in a virtual or in-person workshops where executive leaders come and use so many AI tools that we make available for them that they can actually start to take their own documents and reimagine savings of time and energy but taking away friction from the work that we do each day later on we actually empower different departments, hate your finance sales marketing operations and start to utilize AI across relevant use cases to upskill their different types of departments if you're interested and ready to go on your AI journey we'll be able to showcase over 700 applications of applied AI but how you can actually stop re-eventing the wheel and start using AI for your organization today if you want to start out and find a little bit about the golden thread of activities that we've got allow you to start to utilize AI straight away please feel free to reach out to us and we'll give you a good comprehensive overview of the arts that possible minus the hype purely applied see you then

Podcast Summary

Key Points:

  1. Yealtube specializes in semiconductor manufacturing, focusing on yield, quality, and reliability through advanced data analytics.
  2. The company’s Yealtube Live platform delivers real-time, transparent insights during chip testing by deploying lightweight AI agents on expensive test equipment.
  3. Yealtube Live enables corrective actions in real time, improving first-pass yields and maximizing existing test factory capacity without requiring new hardware or updates.
  4. The platform centralizes fragmented data from manufacturing and testing processes, creating a single source of truth and breaking down historical data silos.
  5. By integrating machine learning and contextual data, the platform predicts equipment failures and recommends actions to improve efficiency and reduce costs.
  6. Yealtube is expanding with a new AI-native platform (Yieldtop AI) that allows customers to build their own AI models using real-world production data.
  7. The solution empowers both humans and AI, fostering upskilling through collaborative learning and preserving institutional knowledge.
  8. Real-time data access and contextual insights enable better decision-making, forecasting, and digital transformation in smart manufacturing environments.

Summary:

Yealtube is a leader in semiconductor manufacturing analytics, offering a data-driven platform focused on yield, quality, and reliability. Its Yealtube Live product deploys lightweight AI agents on high-cost test equipment—without disrupting production—to provide real-time visibility into manufacturing performance. This enables immediate corrective actions, improves first-pass yield, and maximizes existing test capacity, helping companies defer capital expenditures.

The platform centralizes fragmented data from testing and manufacturing, eliminating silos and providing accurate, up-to-date insights. Advanced machine learning models analyze historical and real-time data to predict failures and recommend optimal actions. Yealtube is now expanding into an AI-native ecosystem (Yieldtop AI) that allows customers to develop their own AI models using real-world data.

The solution not only enhances operational efficiency but also fosters human-AI collaboration, enabling workforce upskilling and better decision-making. By making data accessible and contextual, the platform supports smarter, more responsive manufacturing—critical in a high-demand industry where transparency and scalability are key.

FAQs

Yealtube specializes in semiconductor manufacturing, focusing on yield, quality, and reliability through advanced data analytics and AI-driven insights.

Yealtube provides real-time data from test equipment, enabling instant recommendations for corrective actions that boost yield and maximize equipment utilization.

Yealtube Live is a lightweight software agent installed on testers that streams real-time production data to a recommendation system, offering actionable insights to improve yields without disrupting production.

Yes, by identifying issues early and improving yield and capacity, customers can defer investments in new equipment, reducing CAPEX needs.

Yealtube cleanses and centralizes data from various sources, eliminating delays and manual updates to provide reliable, real-time data for AI models and decision-making.

Yealtube offers AI-powered predictive analytics to forecast equipment failures and yield drops, and supports customers in building their own AI models using real-time data.

Chat with AI

Loading...

Pro features

Go deeper with this episode

Unlock creator-grade tools that turn any transcript into show notes and subtitle files.