2025 Bill Gross Business Plan Competition

Good afternoon everyone and welcome to the second annual Bill Gross business plan competition. I'd like to offer a warm welcome to you. My name is Fred Fina. I'm the chief innovation officer at Caltech and I'd like to also offer a warm welcome to the alumni that are joining through the live stream. For the first time we have a live stream. So welcome to you. I'm also an alum like many of you here. So this is the second time we do this competition and I'd like to also add a few words about the various entrepreneurship programs we have at Caltech. Now, it has changed over the last few years. We have several things going including an internal venture fund to support companies coming out of Caltech. We have great entrepreneurs in residence to support entrepreneurs here at Caltech who want to start companies. We have an incubator space now a few hundred yards from here on Del Mar that's housing currently 12 startups out of Caltech. We're working on a new larger building, a 120,000 square foot building on Green Street that will house among other things anchored tenants hopefully in the pharma area and also a life science incubator. Part of these programs include an internship program for students in the summer and one of them is a Bill Gross summer program. Thank you Bill again for supporting this program. And the other one is a Tim Ryan program and both of them support students who want to explore either their own ideas or want to work in a startup during the summer. And so all this ecosystem is growing and of course today's event is the Bill Gross business plan competition. In the last two years we've received over 150 submissions for this incredible novel interesting ideas from students. And in this year alone, we got over 70 submissions. And today you will see presentations from nine finalists. And so it's been pretty amazing to see the outpouring of new ideas from the Caltech students. You know, the culture's changed over the years. You go back 10, 15, 20 years and there wasn't this vibrant atmosphere and excitement for entrepreneurship. So really excited about that. So today we'll have a panel of judges that will be judging the competition and then awarding the winners, the top three, but also the audience and the live streamers will be able to vote for the fan favorite. You will have a QR code and then we'll be allowed to pick your favorite team. So I'm going to stop here. I want to thank again Bill Gross for making this possible and on to you, Bill.

Hey everyone, I'm Moral. Thank you so much for coming out and supporting us for the second business plan competition. We're excited to have you here. As Fred mentioned, you'll be hearing from nine teams today. Each pitch is going to be five minutes long and then it'll be followed by three minutes of Q&A from the judges. And then following the presentations, you guys will hear from past winners and what they've accomplished this past year. And we're doing something new where you can vote for your fan favorite. So, please be sure to stay and vote. And then let's see, after that, we have the reception and the prize announcement. That'll be around 3:50. And if you're a competitor and can't stay for the awards, please let a staff member know. And then also, once again, these pitches are being recorded, so it'll be available on YouTube after the event. And then before we begin, I would like to ask you all to please silence your cell phones. Thank you very much. And then with that, I think we can start. And I would like to welcome up the first team, Natrium Labs, who'll be presenting on their dietary supplement to regulate sodium consumption.

Good afternoon everyone. I'm Yameng Zhang. I'm a grad student in neurobiology at Caltech. My PI and I co-founded Natrium Lab. So this is the first and only company working on sodium appetite and we aim to reduce the high sodium intake spontaneously. So let me highlight a critical but overlooked problem in our daily life. We are consuming far more sodium than we need every day. So here's the data. On a global scale the average sodium intake is twice of what WHO recommended and in the US over 90% of the population exceeds this threshold and high sodium has been causally linked to hypertension, renal failure, and even stroke. So in the US more than 120 million people have high blood pressure and are under elevated risk of cardiovascular disease. So this is why sodium reduction is not just important, it's urgent. At Natrium Labs, we are trying to address these needs with a science-based daily supplemental approach. Despite some proactive health individuals trying to stick to the lower sodium options, the general salt consumption remains persistently high. Currently, there is no way to address the biological drivers behind this preference and craving. And according to our market research, 70% of the consumers are trying to stick to a healthier diet, but they fail to do so because the low sodium diet is not satisfying and they have a strong craving for salty food. So the pain point here is not just access to lower sodium options, it's to make these options more satisfying and more sustainable. So, Natrium Lab is developing a novel neuro-targeted supplement to address this unmet need. This graph illustrates dopamine levels. This is a reward signal. And this dopamine level will vary when high sodium food is consumed under different hormone conditions. So based on our published research and filed patent, we have identified a specific neural pathway that modulates our perception to sodium. And by targeting these molecules and receptors, we can make people crave less sodium food. There are 80 million US adults that are actively seeking to prevent or reduce high blood pressure. Assuming a 5% market penetration rate, this will yield a total market of over two billion dollars per year. And importantly, this $45 per month subscription is trivial compared to a $65,000 procedure due to heart failure. So, our product is designed not to just improve health outcomes, but to do so early, accessibly, and affordably. We have already completed market research and collected rodent data. With the current formula we successfully reduced the sodium intake in mice by half and by July we will finalize our formula and conduct small-scale human tests. So the human test is not required for any supplements but we insist to do so due to the quality of our products and we will start doing marketing and release the product via e-commerce through the partnership with clinical institute and our goal is to launch commercially by early next year and this lays the groundwork for future product lines spun out from our lab based on the same platform. We are currently seeking seed funding and these include SBIR grants and we are also talking to Caltech internal seed fund and funds will be used mostly in the manufacturing and marketing sections. What makes Natrium unique is our science-driven formulation and our strong team with deep expertise in the neuroscience of taste biology and translational research. We are also recruiting more doctors from UCLA to bring more expertise in the cardiovascular field. At Natrium Lab, we believe the future of preventive health is not just about restriction. It's about changing how the body perceives and experiences food. So, we combine neuroscience with consumer health. And we create a new kind of supplement that helps people eat better naturally and sustainably. And I want to thank my mentor Will and also thank Jay and Isabelle and the OTT office and thank Bill for this chance. And thank everyone for your time. We are excited to bring this vision to life with the right partner. Thanks.

Thank you very much Yameng for highlighting this very important problem. I liked some of the things that you said around really taking a scientific and specific focus. I guess some of the questions that I have are predicated on a couple things. So the first is did I understand you correctly where the human test is not required for a supplement but it's just something that's there to confer additional validation. Is that right? So if so is the idea that you're looking at the direct to consumer channel is that how you want to actually market the supplement?

So we are targeting hormone level and the receptor. So the interesting thing about this idea is what we are targeting is outside the blood brain barrier. So the molecule can directly act on this and so the compounds we are using for the supplements they are already safe approved and we can find it with the right manufacturer. So yeah.

Do you intend to sell directly to consumers?

Yes we are thinking about that. And then the second step maybe B2B, it's hard to directly sell to the business or clinics. Yeah.

And do you have a sense of what your manufacturing costs are going to look like and has that started? I mean is this a GMP question, a GLP question eventually or is it pretty verified manufacturing process verified all to the manufacturers?

Okay. Thank you.

Thanks. Yeah great great presentation and it's a great problem right. I may have missed it. Can you just expand a little bit on what this supplement is? Is it something you mix in water? Tell us a little bit more about that and if at all possible how you're modulating this hormonal system that you just referenced.

Yeah. So it's a great question. So in that paper we identified a neuron pathway that can modulate proposed taste perception to the sodium. So there are already molecules proved to downregulate that hormone by the current product and we test screened all the molecules and mixed in a new formula and we tested the formulation itself is much more efficient than the specific ingredients. And with this we want to do capsule but tea bag is also an option. Yeah.

Thank you. And so the levels at which these supplements would be found in your product, do you think have you thought about whether they would require regulatory sort of scrutiny because you're maybe enhancing one component more over another.

Yes. So I think the most strict regulation part is the manufacturer and we will outsource to the good standard like manufacturer source. So yeah, it won't be a big problem for the supplements for regulatory. Yeah.

Great. Okay. Thank you. Thank you so much.

All right. Next up, I'd like to invite Matterware and their AI-assisted chip design software.

Hi everyone. We are Matterware. My name is Walang Guses. My co-founder's name is Robert George and we are both PhD students at Caltech. I'm in a chip design group in Ali's group and Robert is in the machine learning group and we are here to pursue the idea for AI for chip design. As you may all know software keeps accelerating or keeps going at an accelerated pace and generative AI and the current machine learning products are the next paradigm in the software evolution. While that's happening in the software industry, we believe that the software for chip design has mainly stagnated over the course of the last 20 years. And we believe at Caltech we are in a unique position and have the right timing to allow chip design software to catch up to this evolution. One brief example to kind of emphasize this stagnation issue. If you look at the normal software for software development and other workflows, we went from basically automation that went from just giving hardcoded answers and whatnot to complete bots that can do the workflows for you. For chip design, workflows have remained mainly the same. You can see on the bottom a main software for chip design, Cadence Virtuoso, the front end and the back end processes have mainly remained the same. And this basically performance efficiency gap is clearly visible in the cost incurred in the advanced nodes in chip design at the process nodes from like 2010 to 2025 the software has started to incur more than 60% of the costs in a single chip. So we believe this basically benefited greatly the current incumbents where their revenues have increased exponentially over the course of the last 10 years. And these giants are Synopsys and Cadence and we believe these new tools bring the right time to disrupt this duopoly. And after doing a brief market research and talking to some of the executives and engineers at chip companies we see there is a demand that basically we can bring these novel tools and try to automate a lot of the tedious tasks in chip design workflows. If you look at from the top down view our market size in general semiconductor sales is around $600 billion dollars of sales annually and then 50 billion of that goes into chip design software. We hope to penetrate around 2% of that market to get to around 100 million annual recurring revenue. And from the bottom up approach, we have a plan to gain a beachhead in the market by charging 10 times less than what the big EDA companies charge to big chip companies. And then we're going to start by targeting small chip startups who cannot afford to pay these high costs. And with this we aim to also get to the revenue mark that we want to achieve. So this is our software briefly in action. On the left you can see a chip designer colleague of mine who's trying to design a photonic chip component. You can see the progress bars for this design on the top. On the right we have our own software that can autonomously design this component 10 times faster than what the chip designer could. And with this we hope to basically achieve 10x faster chip design with 10x lower cost. So that we can democratize chip design software not from big chip companies to all chip startups. Our differentiation is we want to basically leverage the expertise in two very deeply specialized fields. Kind of combining our expertise with chip design and machine learning and also leveraging the basically large and diverse chip design IP at Caltech with permission from Caltech. We believe we can train models that can outperform the other products by these big EDA giants in the long run. Our milestones are basically with an hopefully initial prize money investment. We want to roll out a product that can aid chip designers in their chip design workflows and we want to get to 10 customers to hopefully 100k ARR and recurrent revenue and then from this revenue we want to scale to a point where we start to train our own models and roll out hopefully foundation models for chip design and start competing with big EDA giants. So this brings us to an opportunity to invest in bringing some of the new software tools and the revolution to chip design software. We believe we're at the right time and unique position to bring Caltech IP and expertise into this domain and we believe we have the expertise in two very specialized fields that many other companies do not. And with this we hope to disrupt the duopoly that's available in the market and democratize chip design software. Thank you for your attention. If you have any questions we can take them.

Thank you for the presentation and very timely, this is a very timely presentation. So thank you. You mentioned the incumbents but there's definitely some startups. Are there any startups that you have in your mind that you're looking out for? And my second question is you mentioned that with Caltech IP you can use that information to train the model. Is there any other type of sources that you use and how do you capture all the knowledge that is in the people's heads? I forget the word for that. That is one of the big problems for these chip design companies to actually train models.

Yeah absolutely thank you for the question. Yeah there are a few startups that have come out in the last few years. I know a few like Silimate and similar stuff in Y Combinator. We believe that their focus is on certain chips that are in the digital design and sort of verification side of things. We believe we also have IP in the RFIC and photonic chip side of things that those companies do not. And the IP at Caltech particularly excels in this because there is a lot of research going on in those domains. So we want to differentiate ourselves from those startups by targeting those verticals in the chip industry. If what was the second question if I missed that?

Yeah, the second one was about how do you capture tribal knowledge that is one of the big parts of chip design.

Yeah. So I think as a start we want to leverage the current research and the IP that's happening at Caltech but in the long run we hope to also basically gather more data and measurements by having tape-outs of chips and measuring them so that we can feed more data into our models. That also brings us a good opportunity to turn basically this idea into sort of an IP company as well. So we can verify our designs and also license our IP to big chip companies in the long run.

Thank you very much. I love the idea of disrupting the EDA duopoly, right? That's wonderful. One of the things that I was curious about is are you in a position with your early model, the early model is a subscription model, right? I think you said 10,000, right? But are you in a position to actually start to capture more of the value that you're creating? Because the concern that I might have and I'd love to see how you guys are thinking about this is all of a sudden you have a recurring subscription model, right? It's very easy for companies to get locked into I'm only going to pay this much because this is the value that I see that's being created, right? So any thoughts on how that relationship can change so that you're able to extract more value?

Yeah. So actually the current negotiations and the pricing model for the big EDA giants kind of go on. There's not a set price in their subscriptions. There's usually like a negotiation that happens within the chip design company and the EDA company. So we can follow a similar kind of structure based on the needs that the chip company has. We can adjust their licensing fee based on how many designs, how many tape-outs they do per year. But the one thing that we want to still differentiate ourselves with we believe that there's a lot of secrecy in basically setting the prices for these EDA software. We want to kind of put some relief onto that secrecy and put our price up front so that people don't need to negotiate and go to these channels. But I think the final solution is going to be having open negotiation and putting prices on our website and everything throughout the process. Thank you.

Okay. Thank you so much, Matterware. All right. Next up is Nexa Labs and they'll be presenting on their minimally invasive microchips for veterinary health monitoring.

All right. So, hey everyone. Thanks a ton for coming. My name is Zarif and I'm co-founder of Nexa. So we want to empower farmers to understand, manage, and then capitalize on the health of their livestock because look, globally right now, farmers are losing about $400 billion to livestock health issues. And when I say livestock right now, we're talking mostly about cattle. So just like us humans get sick, cattle get sick, they spread disease to one another, they grow less, they die, and in the process, farmers are losing money. Now, today we've talked to about 70 farmers all across the US, and we've learned that the vast majority, they monitor their animals' health through a manual subjective process where they pretty much just kind of look at cows to see if they're sick. That takes a lot of time. And as you can imagine, it doesn't actually catch health issues before they start spreading and devastating these large herds. And existing technology here is not much of a step up. They're these external wearable collars and ear tags, but they sit outside thick animal skin, so they can't actually collect useful data or deliver meaningful insights. There's also these sort of smart pills, but these have these large batteries that cows are supposed to swallow. And from the farmers we've talked to, we've learned that folks don't want to do that, and it causes regulatory hurdles. So our solution then is a comprehensive livestock health monitoring system with minimally invasive hardware that tracks the health of cattle in near real time. We want to turn that health data into meaningful insights for farmers like early disease detection or telling a farmer, hey this cow family is doing healthy. You should breed it more or hey this family is consistently sick. Maybe you shouldn't breed it anymore. Now critically the whole process should have a low burden of adoption so farmers can scale up the whole system quickly. So, what we've been building out at NEXA centers on an injectable microchip, a small device about 4x50 mm that can fit inside a standard animal syringe and be injected at the base of cattle ear just underneath their skin. Now, keep in mind that cattle already get injected with vaccines about twice per year. This microchip collects animal health data near real time, sends it to a small module just outside the ear. That module can take the data, send it to the cloud and we can start delivering value for farmers from day one. So our initial approach is to go after early disease detection. For example, bovine respiratory disease which is the costliest and most common cattle health issue. We estimate that just with the first version of our product, we can increase profits for an average large cattle operation by up to 15%. In the future, we'll go after more animals like horses and pets as well as increase the range of diseases that we can detect. And so this is a massive market. Even just looking at cattle operations here in the US, we're looking at about a $10 billion market. And in the first few years, we think we can do revenue of about $100 million give or take. As we expand to Canada, Mexico, and internationally, and expand the number of animals that we can track, those numbers will obviously only increase. And so we are delivering value to farmers by saving them time and money so we can charge hardware and recurring software costs for that. But with our technology, we're also going to be collecting the world's largest livestock health data set, which we can monetize in many ways. We can sell the data to cattle insurers so they can offer farmers flexible rates. We can sell it to livestock pharma so that they can do clinical trials of their therapeutics better, so they can deliver better antibiotics and treatments for cattle. So once we have this data, the possibilities are endless. So far, we've built out early prototypes of our product. We're patent pending and we have an agreement with Cal Poly Pomona to test our device on their cattle this upcoming summer 2025. We're also proud to have already partnered with some awesome investors who've helped us give us some activation capital if you will from the fall once we validated our device. We're going to start our go to market with a regional network-based approach and that'll take us through this fall and early into next year. So look, we're a diverse team, but our backgrounds are in biology, AI, and hardware at some of the nation's top research labs and startups. We're also advised by Dr. Gakara, the head of Cal Poly's livestock research institute. And so look, the current players in the space, they're these slow-moving companies that are going for the technical low-hanging fruit, these wearable external monitors. We're from Caltech. We can move fast, iterate our technology rapidly, and actually deliver solutions that matter. We've talked to farmers. We know that the current tech doesn't cut it. We know that they have low adoption. So, we can be poised to bring a product that can actually make an impact and capture a large percentage of the market. So, today, as I mentioned, we're already backed by some investors who are really excited to have on our cap table. They've backed some of the leading deep tech companies out there like Northwood, K2, Varta Space, and more. We're seeking to round out this activation fundraise so we can do our trials and start our go to market. For example, we want to get out to as many livestock conferences as we can to get in front of as many farmers as we can. We hope you're as excited about NEXA as we are and we'd love to answer any questions. Thanks.

Judges. Do you have any questions? Yeah. Great energy. Thanks for it. I want to hear more about it. My question is when talking to the farmers, how much of the kinds of things that you guys would be detecting in the cattle are things that they would treat? I suspect that there are some aspects, some kind of problems that would just lead to the animal sacrifice. So I'm wondering if you might be able to help them with that.

Yeah. Yeah. Of course. So here's the deal right when a cow gets sick a farmer essentially has two options right? One is you treat the animal. So we're talking about bovine respiratory disease, your sort of standard antibiotic, it can range from $50 to $150. So a farmer can either treat them or as you mentioned just kill off the animal. Now when you're at a small scale of disease, like let's say two out of your 1,000 cattle are sick, that's fine. You can just kill them off. You don't want to treat them. The issue comes when the disease starts spreading to 5, 10, 15, 20, 30% of your herd. Then if you kill off 30% of your herd, you're losing a bunch of money. If you go to spend the money to treat all of them, you're also losing a bunch of money. And so that is where our tech really brings value. If you can detect disease earlier, when it's at two cattle instead of 30% of your herd, you can either just kill them off or treat them, do whatever you want. The issue comes when it spreads. Just for some context on now the data on this is not so good because cattle health data is not collected widely but on your very much averaged out sort of cattle operation in the US just talking about bovine respiratory disease somewhere between five and 45% of cattle will be actively infected. Call it in the middle let's say it's 20%. Even that is a daily problem that farmers deal with. It's not like an earthquake insurance where it's once every three years something goes wrong. This is an active problem that folks deal with and that's what we've come to learn from talking to them.

Quick question on product roadmap and evolution. So this goes on what sorry oh product roadmap and product evolution. So as I understand it this is technology that's come out of Wei's lab right. So there's sorry no sorry please go ahead.

No, I was going to say just to be clear, we are a team of undergrads. We've built this on our own. We've taken no Caltech resources or funding. We're not Caltech IP.

Gotcha. Thank you for clarifying. Yeah. How do you see the what do you think the cost will be and how long does it last in the cow? Like do they have to be replacing it?

Yeah, sure thing. So, right now the cost so I'll give some context. The cost to manufacture on the devices we're talking about right now between $10 and $12. We think once we do sort of mass manufacturing we can get that lower to more like 8, 9, 10. We're planning to sell at 25 to 30 for the hardware unit. For context the ear tags and collars those range from about $30 to $80. So we're going to be undercutting or at the lower end of that. Yeah. Sorry, did I answer your question or was there another?

Yeah. And how long does it last?

Oh yeah, yeah, sure. So first of all, we can't tell you for sure because we're still developing the product. But what I can tell you is the number one issue is if the device is in the cattle and the device itself harms the cattle after being in there for a year, that would be a big no no. But luckily, our device is encapsulated in PDMS, which is used often in human implants. And that we're confident will last about three, four years, which is pretty much the lifespan of most of the cattle that we've been going across. So, we're confident that we'll have to sort of get the device to last as long as it can, but it's not going to harm cattle even if it stops working.

I think we're out of time now. Thank you so much.

So, I'd like to welcome Mind Mesh up and they'll be presenting on their laptop downtime utilization to supplement AI processing power. Please join us.

More than half of the world's GPUs are idle at any given time. That's not just an inefficiency. That's a massive untapped market. And it raises two big questions. How do we harness this lost potential? And who do we empower it with? We're Mind Mesh and we're here to give you some answers. Today's computing infrastructure is fundamentally exclusionary with all the focus being on centralized servers which charge thousands just to train a single model and that locks out startups and researchers as well as all the untapped power in their personal devices. And this is where the decentralized computing market comes in where anyone can rent GPU access by connecting remotely to a device. And this is the stage where Mind Mesh plans to deploy its mission. What we're proposing today are two things. Firstly, a vision, a world where decentralized computing can power individual innovation by granting you access to industry computing resources without the need of a data center budget. And secondly, a concrete step towards that solution. We're proposing a novel smart algorithm that can drastically improve upon the efficiency of existing networks and truly make democratized AI a reality. Secure, scalable, and efficient. This is Mind Mesh. So basically, we're collecting a bunch of GPUs worldwide. These are our nodes. And basically we're creating a unified system, a blockchain system whereby all of these nodes can collaborate together for fast data processing. And here is how we're very innovative. First, we have a smart task splitting schedule which is an AI-based scheduler and what it does is that it scans the whole task and basically it splits it to multiple jobs. It sends them to multiple GPUs and that makes the process way faster. Second, we have an AI-driven pipeline in which basically overlaps task stages like data cleaning, preparation, training and more. And we also have the parallel code execution feature which means that for the most important code basically what we do is that we send it to multiple nodes and then the fastest node to actually finish the job reports back to the recipient. We also have other features such as the regional clustering feature and the federated synchronization trees and also they allow for faster data processing. Now, on the security front, we've got you covered because first we have ironclad encryption, which means all of the data is held tight. And we also have zero knowledge proofs, which means the owner themselves of that node doesn't really know what is running on their GPU, which is kind of crazy, but it's amazing. Now, this is the right time to act. Why? Well, first, internet is way faster right now. Second, AI is booming. And because of that basically we have a $50 billion market niche. So basically the current cloud

Providers such as AWS are not able to meet efficiently and affordably these needs. Now, if you look at our direct competitors, which are also startups that focus on decentralized computing, then you're going to find that these startups don't really match you with more of one or two nodes and cannot handle large-scale projects. And here is our pricing. So, basically, we charge per hour of use. So, $2 for top-tier nodes, 50 cents for mid-level nodes, and five cents for entry nodes. And we also charge for premium features. We plan to gain exposure through pilot programs with research labs and startups as well as with a clean web interface for individual access. And we will reinvest our revenues into enhancing this API but also in improving our efficiency and scale by onboarding more global nodes. We're basically asking you for a seed investment of $30,000 and it's going to be allocated as follows. Well, 40% is going to go for the development of our MVP. 30% is going to go for our beta program. 20% is going to go for adding more nodes so that we have a bigger network and 10% is going to go for documenting our results so that we talk to more sponsors and get more funds. Mind Mesh. Invest in a future where innovation is accessible to everyone, not just the few. Thank you.

Judges, do you have any questions?

Yeah, so I guess here the key differentiator is that other people also thought of decentralized computing but what they do is that basically they match you with only one or maximum of two nodes that match your needs. For example, I go to their website, I just put in my task and what they do is that they have a smart contract that it's going to allocate your task to one or two nodes and they need to be in the same proximity. The problem is that this has not been very successful because research labs and startups need processing power for larger projects and this idea basically solves this problem because here we're allocating the task to many nodes and here basically all of that comes in the implementation of the smart contract. So what we're doing is that on a side we're using Python to develop an AI-based algorithm that scans the code first, see if it's parallelizable, meaning you can divide it in multiple batches, and second, can we use pipelining on that code? For example, if it's TensorFlow-based, it can be divided on a pipeline, meaning one for data cleaning, one for preparation, one for computing, one for training, and this saves a lot of time. And also basically we have the regional clustering feature which means that we're not allocating nodes only on node capability because otherwise you can have one in Japan and one in New York and this is going to make you lose a lot of time. So what we do is that all of these we group them together and we use the federated synchronization tree which also AI-based means that when these nodes talk to each other about the progress they made on a certain task they're not going to basically report every single detail. So we have another mini algorithm that's going to shrink the data to be transferred and only the most important data is transferred between these nodes. They group everything together and then it's transferred to the recipient. Thank you.

That's time. Thank you so much. Before we invite our next team up, one quick housekeeping note for the judges. If I could please ask you to speak more or project more into the microphone, that would be great. Thank you. All right. Next up, we have Echol Bloom or Echol that's going to be presenting on their platform for early autism spectrum disorder detection.

Hi everyone. I'm Mila. And I'm Samia. And we're the founders of Echolue, an AI-powered ASD, autism spectrum disorder onboarding assistant that provides insight that grows with you. Supporting children with ASD requires deep individualized understanding. But critical knowledge is often trapped in scattered documents, unstructured notes, or the memories of experienced staff. Onboarding time is time-consuming and information is fragmented across documentation leading to missed insights and low continuity of care especially as staff turnover. Introducing Echolue. Insight that grows with you. Echolue is an AI-powered onboarding assistant that synthesizes ASD-related documentation like diagnostic assessments, educator notes, individualized education plans, and caregiver input into a dynamic personalized profile of a child's triggers, interests, strategies, and milestones. As caregivers, parents, and educators upload new documents, the profile evolves and grows with you. Echolue also synthesizes all this documentation and even video into actionable insight that users can provide feedback on. It instantly integrates all this information into a child's milestone dashboard, flags delays, and suggests personalized strategies all backed up by caregiver evidence. Now my co-founder Samuel will go into the model architecture.