OpenAI Academy x GitLab Foundation: AI for Economic Opportunity Demo Day

[00:00:00]

I'm Ellie Bertani. I am the president and CEO of GitLab Foundation, and I am just thrilled to invite you all here in partnership with OpenAI to what I assume you will find to be an incredible exhibition of amazing grantee projects today at our second annual Demo Day. We're here to celebrate 15 incredibly innovative organizations who are going to show you the work they've been doing over the past eight months to drive economic mobility in sometimes really unexpected ways. We're going to show the power of what's possible when we focus the power of AI on social innovation and on economic opportunity.

I'm so grateful for you all taking the time to be with us here today. And in a moment, I'll be kicking this off alongside my friend and board chair, Caroline Whistler. But before that, I have a whole set of people I'd like to thank for helping us make this event come to life. First, I want to thank our hosts and partners, OpenAI and the OpenAI Academy, who have played such an integral role in getting the AI for Economic Opportunity Fund off the ground over the past few years.

We're grateful to be here in this incredible space today and for your partnership, OpenAI team, and the investment of the time you spent and resources on this very special group of grantees. Thank you especially to Alex Nawar, the head of OpenAI Academy, who has worked so closely with the grantee organizations, and also to a group that often we don't get to recognize, the group of engineers that have been supporting these teams over the past eight months. So Yogi, Victor, Anika, Charles, Mihei, and Joshua, if you're in the room or if not, if you're live streaming alongside with us, these teams wouldn't have been able to accomplish what they have without your support, guidance, and time.

[00:02:00]

Thank you also to Chan Park, who we're going to hear from in a few minutes representing OpenAI. He's going to share some opening remarks that I think you're all going to find very interesting. And thanks to the many other members of the OpenAI team who have helped us set up the venue, prepare assiduously for our time today, and really help bring this event to life. Thank you also to our friends and partners at the Annie E. Casey Foundation, in particular Elizabeth Nash, who's here with us today, and to our colleagues at the Ballmer Group, particularly Kevin Bromer, Corey Klein, and Soon Lee, who invested so much time and resources above and beyond our own to help bring our grantee projects to life.

And finally, thank you to the GitLab Foundation team. So many of you have put in so much time into this work. Matt, Sammy, Alicia, Emily, Allison, Alex, CJ, none of this would be possible without your hard work. So thank you, team. This is our third AI cohort and our second AI for Economic Opportunity Demo Day. Gatherings like this are important because they show what is possible when we really come together with a spirit of collective action. The collaborative spirit is even more important given the pace of change in society today, the significance of the problems that we faced as a society, and the creativity of the social entrepreneurs around us.

The organizations that we'll hear from today represent some of the most diverse and broad uses of AI that we've seen across our three years of doing this work. We have solutions that range from improving the day-to-day work of African pharmacists to helping farmers worldwide manage the effects of climate change to supporting frontline workers negotiate wages in a time of growing economic disruption. GitLab Foundation's North Star

[00:04:00]

to drive at least $100 of lifetime earnings for every dollar that we invest. The projects that we will hear from today collectively are estimated to generate roughly $1.4 billion in additional lifetime earnings for only $4 million invested. The 15 projects are expected to touch over 27,000 people, increasing their lifetime earnings on average by about $52,000 per person. And these innovations are just getting started. It's truly an incredible representation of what is possible if we move quickly and leverage the most powerful technology on the market for good.

Three years into this fund, across more than 1,400 ideas submitted, we can see that the ambition across the social sector to leverage AI is tremendous. Yet, across philanthropy, our funding system has been challenged to keep up with this potential. In each of our three funds, we've serviced probably 50 to 100 fundable projects, but we only have the resources to fund about 15. I feel strongly that we as philanthropic funders collectively need to accelerate our work and invest substantially more in order to adjust the challenges that face us as a society.

Many promising concepts go unfunded or get stuck in the pilot phase because we don't move quickly enough or apply enough risk capital in this space. I spent more than 10 years of my career in the private sector at some of the biggest companies of the world, and what I saw there was that every day our colleagues in the private sector make enormous high-stakes decisions based on limited data. They take calculated risks to drive value with a high sense of urgency. We should be doing the same.

One stat that has stuck with me and I've cited many times over the past few days, in 10 of the largest venture-backed AI funding rounds

[00:06:00]

2025 alone, $84 billion were invested in the private sector development of AI tools. In comparison, estimates are that $2 to $4 billion in that same time period were invested in the social sector. That's a 20 to 40x difference in private versus social investment. And this mismatch is distorting the role that AI is playing in our society. We need to move faster to ensure that more resources are brought to the brightest leaders working for social good, and that private markets don't fully define the architecture of our society's future.

And I think we owe it to the social sector so that we all don't get left behind. And so today is a day to inspire us, to give you in this room just a little flavor of the most exciting, most innovative work that AI can unlock for economic opportunity. But remember, these 15 ideas and organizations, behind them there are dozens, hundreds, likely thousands more that could and should be resourced to do similar work. I hope that today inspires the funders in this room to invest in these specific ideas and organizations, but that it also encourages you to accelerate your funding beyond this day and beyond this room.

There is so much that we can do together to drive incredible outcomes. Now I'd like to invite my friend and colleague Caroline Whistler to say a few more words about the day. Thanks, Ellie. I'm Caroline Whistler. I'm chair of GitLab Foundation's board. I'm so grateful for you all today and for all the work that went into having this demo debut possible. GitLab Foundation was founded four years ago, almost to the day. And it was founded right around the same time that

[00:08:00]

OpenAI GPT was made available to consumers, which I think is both unique and a testament to how we live and operate as a foundation. So our operating foundation, operating model as a foundation is really built around the belief that as a foundation built in the age of AI, we need to use the full power of AI to serve our grantees, with a singular focus on our North Star, which is raising people's lifetime earnings through access to opportunity. And for us, what that looks like is deploying AI in a variety of ways, including combining rigorous pre-investment ROI modeling.

You've heard about our ROI. If you haven't, you will. And also our AI-enabled diligence. And as Ellie talked about that, talked about what that really allows us to do as a foundation, even though we're four years old, I believe is to help move with speed, which is so important in this moment. So GitLab Foundation has also set a North Star for ourselves and that we want to see 100x impact return on investment for every dollar that we put in as a foundation.

We then, we use that ROI to have, I've really set that North Star, but then have the discipline to measure it rigorously, interrogate it, and to learn and adapt more quickly over time. And that has helped us drive some incredible results just over the past four years. Our return on investment for about 70 grants that have closed out over these four years collectively is 232x. Meaning that for every dollar we've invested, we are seeing 232 times that in economic impact, in increased lifetime earnings.

That is billions of dollars that may not otherwise be there for people across the US and the world in our grantees. Achieving that kind of impact requires

[00:10:00]

requires hard work, rigor, and also incredible partners, which is you all and what our Demo Day reflects. Today, the foundation sees the role of philanthropy as truly evolving. Working together is more important than ever, and for many who want to give today, the experience of giving has never felt more consequential in this moment. But there's a significant amount of uncertainty, and so sometimes capital sits on the sidelines, not for lack of will, but maybe for lack of infrastructure, maybe for lack of trusted relationships to move it well.

Partnerships are essential to getting that capital off the sidelines and serving grantees, like the ones you all hear from today. We increase confidence, and we would say we increase outcomes when we work together, pool efforts, and nurture the best ideas. Our goal in philanthropy should always be to move at the speed of the world around us, to be responsive to what our communities need to thrive, and to catalyze the future that we want to see. We do that by working together.

The pace of change demands it. So today, you're going to hear directly from 15 incredible project teams that are truly building the future we want to see with AI. Ellie talked a little bit about them, and we are eager to connect them with all of you. Funders, potential partners, collaborators in the space, so that we can build an ecosystem that moves with speed, responsiveness, and drives the most impact. Thank you all, and I'll give it back to Ellie to introduce our first speaker.

Thank you, Caroline. Okay, before we get into our pitches and presentations, I'm so excited to welcome Chan Park, head of US and Canada Policy and Partnerships for OpenAI.

[00:12:00]

to give an opening keynote. In his role, Chan oversees OpenAI's engagement with federal, state, and local governments, as well as third-party groups and partnerships. Before joining OpenAI, he led congressional affairs at Microsoft and had a long career in public service, including serving as the general counsel for the US Senate Judiciary Committee under Senators Patrick Leahy and Dianne Feinstein. Before that, he prosecuted a range of federal crimes as an assistant US attorney for the District of Maryland. Chan also co-founded the Asian Pacific American Legal Resources Center here in Washington, DC, and I'm excited to welcome him to the stage here.

Chan, thank you. Okay, I was not told this was a keynote, so. I have some remarks, and most importantly, I wanted to welcome you all to this event, the Demo Day. On behalf of OpenAI, we are really, really excited just to have you all here today, and more importantly, just excited to hear the pitches, see the demos, and hear from the builders in the room. Before getting to some of the written remarks, who are the folks here who have built on AI tools?

If you can raise your hand. Doesn't have to be ChatGPT or OpenAI's tool, doesn't matter at all. Okay, who are the funders in the room? Okay, the builders, keep an eye out for the funders. The funders keep an eye out for the builders, and vice versa, because I know there are a lot of funders here who are also builders, and builders who will become funders. And that's part of the point of today, and I really, really want to thank the GitLab Foundation, want to thank the Ballmer Group, the Annie Casey Foundation for your partnership in pulling people together and in having these demo days.

I think we are at number two, right? And hopefully we'll be at number four,

[00:14:00]

5, 10, 15, and every year, the more and more people we have here who are alumni, who come to talk about the tools that they've built, and that they're implementing and deploying, it will inspire more and more people. And that's our hope, is that we can continue to be part of that partnership. We are extremely proud to be working here as part of this broader Demo Day effort and the months- long and years-long effort. Ellie mentioned some of the folks who've been providing technical assistance over that time to the builders, also the API credits and all the resources.

That's our privilege to, frankly, be a part of that. We know that the folks in the room who are building these tools are on the cutting edge of the technology, on the cutting edge of bringing social good and change to communities using these tools. We're proud to be on the cutting edge of AI technology. We're proud to be on the cutting edge of building the AI models that we hope will really be useful to folks. But you all are on the cutting edge of making sure that it comes to life and actually brings benefit to the communities.

Our mission is to build tools and to ensure that AI benefits all of humanity, but you all actually help bring that mission to life. And so we really honor that and are appreciative and grateful for the work that you do day in and day out. And honestly, we're learning from you, the ways in which you use the tools, the ways in which, frankly, you give us feedback and say, hey, this is not working so great, or this is useful in this way or that way.

That's helping us to make the tools better and helping us to, frankly, fulfill our mission. So we're grateful to you for that as well. The mission put into practice is to make sure that the AI benefits reach communities that are historically less empowered or have less access to technology, capital, and opportunity. That comes in different ways and different forms. It could be in the way

[00:16:00]

Workforce space, it could be an education, could be an economic mobility, health, access to essential services. But ultimately, it is about making social impact and social impact for communities in a way that is designed by and with communities, not just delivered to communities. I think that's been the real beauty that I've seen walking around the different demos, is it's designed with and for communities and with the understanding of what the community actually needs, not just something that pops into your mind, you build it, and then you drop it into a community and think maybe it'll work.

We'll see, we'll find out. Who knows? We know also from building our tools that the way we try to do things is through iterative deployment, right? By building something, getting feedback, making it better over time, working with the communities, the builders, the developers. You're doing the same exact sort of thing in your communities and with the different folks that you're helping to deliver these services to, but doing it with those community members. And I think kudos to the funders in the room as well, because I think my hope, my understanding is that you're also doing this with the grantees, with the builders in the room, kind of hand in hand on this journey to make sure that's not something where you just sign a check, drop it, and then let it go.

The speed of funding, the speed with which we need to move, that's all incredibly important. We had a conversation earlier today about the agility that's needed in this moment, the agility of both the nonprofits, the builders, the funders, all to make sure that you're working together to ensure that you are building the tools, but building them on a timeline that actually will make a difference in the time you needed to make a difference. We all understand and I think appreciate there's real potential here for AI to make an incredibly important, powerful difference in people's lives in so many different ways, whether it's a healthcare context or the

[00:18:00]

workforce context, just making nonprofits more effective and more efficient or just delivering services in a really, really important way. We also know that there are downsides potentially. We also know that there's a real responsibility on us as OpenAI, as developers of these tools, to ensure that they have the right guardrails and safety embedded within it and that we're helping you also to deploy safely. And we also know that there's really an imperative to make sure that people understand what the tools are and what they're not, how they can help and how they may, frankly, be harmful, and making sure that we all are responsible in that way.

And so that's our shared responsibility and our shared commitment to you to make sure that we're partners with you along that way. As I said earlier, you all are on the cutting edge of developing these tools and providing social impact. This is one of my favorite kinds of events, to be honest. I spend a lot of my days talking to folks here in D.C., around the country, sometimes even this morning I spoke with a foreign official talking about policy, talking about the ways in which we need to make sure that we have the right governance in place.

That's important. What's more important in my mind sometimes, or at least more inspiring, is really seeing the ways in which you're bringing these tools to life and showing the ways in which you can have a real impact on people. Alex Nawar, who's up in the front row, who's helped to build out these OpenAI academies, Anna back there as well, the other people from OpenAI, they get to do the real good, fun stuff of going out into communities and working to both build AI literacy, but then also figure out ways to have...

We did a non-profit jam to work with 1,000 different non-profits around 10 different cities in the United States. That's fun, and that's important, and that's cool and exciting to see. I was able to go down to a national

[00:20:00]

It's so glad to see scientists sitting around kicking the tires of our models, figuring out how to do new ways of material science and chemistry, biology, ways in which you can really, again, make a difference in people's lives in the short term and the long term. So I applaud you. I really welcome you. I hope that you will continue to see us as trusted partners, as responsible partners, and to really hold us to account to make sure that we're building with you in community.

So thank you again. Welcome to you. I'm looking forward to hearing the pitches and seeing all the demos. Thank you. I forgot to introduce Matt. Thank you, Chan. No problem. Oh, it's so good to be with you all today. Really, really pleased at how this all come together in just such a nice full room. So I did a little back of the napkin math on who you're going to hear from today. These organizations are representing well over 50 million people that they serve, the projects and the systems that are working within them.

From millions of farmers receiving better weather services to millions of renters, students, job seekers, parents, workers, public benefits recipients. And it's a very conservative estimate of the systems that these are operating within. There are networks like community colleges, state workforce agencies, high schools, benefits offices, again, worker retraining organizations, and communities of thousands of domestic workers all throughout the country. And there's one way to hear these projects today, and that is around their goals, their milestones, their progress, to see it as a very kind of cynical funder about the challenges about these organizations' progress to scale.

But there's another way I want to ask you to see this today as you listen and you kind of just hear their progress, is that it's a glimpse into the future. It's a future where this is not uncommon, where these projects are fully implemented, where these systems

[00:22:00]

work in new ways, in ways that maybe even are a little more human. And I think that's where we're going, right? We're going to start to kind of see the progress of these models and these products in a very short amount of time, even that we've seen in the last few years, transform entire systems, transform entire ways of operating within our world. And I want you to look for that opportunity, see that on the horizon. So as we get started, I want to hear, just share a little bit of data.

So it's often nice to talk about kind of what's coming, and I want to talk about what actually happened in the last couple cohorts, just briefly. It's only been 18 months about since the first cohort finished, since we're on a one-year grant cycle. And the outcomes data from just those early 12 organizations, there are a quarter million people that have been served with direct real dollars in their pockets of over $1.3 billion already measured. That's not predicted data. It's actual data reported by those individuals, real information validated by our team.

Just a relatively small amount of investment for us, there were only a few million dollars. Those early grants were only $100,000 grants to get started prototyping, testing. Those organizations have gone on to raise over $30 million, largely from you, from other folks that have attended demo days, specifically attributed to these days. So we're grateful that you're here, and we hope that you'll dive into these organizations and really get to know them. So why are we doing this work? Again, we are an economic mobility funder.

We are not a tech funder, but we backed into this because we see the opportunity here. So there's been an endless debate about how AI might impact all these systems, but particularly, of course, jobs, right? People think all the time about the risk of economic opportunity, the risk in education and attention, job quality, job loss, all of this. Real concerns. But in the face of all that uncertainty, right, which there is plenty to go around, we think the best way to go forward is just to act.

[00:24:00]

to give experimentation funding, evidence-building funding, and a pathway to figure out what's a better way to start to glimpse that future for these partners. So in education access and job placement and safety nets and legal counsel, support for low-income workers, and all the other things you're going to hear today, working parents, domestic workers, so many opportunities for upside in this work. And we're really inviting you to join us in that. So this is how this day is going to go. I'm going to get out of the way, I promise you, in just a second.

You're going to hear from the first eight teams in a bit. They're going to take about five minutes apiece just to walk through progress to date, a little bit about their problem, and really what they're doing about it, really spend most of the time actually demoing the progress that they've made. Then we're going to have a brief fireside chat to actually hear from grantees from the last couple cohorts about the progress they've made, impacts that they've seen, just to give you, again, that glimpse of the future of where these projects are headed.

And then we're going to wrap up with the final seven. So let's get started first with the Development and Innovation Lab. Hello, everybody. So today we're going to talk about three things. The first one is the economic challenges facing farmers in the developing world. The second is our ambition. And the third is how AIs can help solve that. So I would like you to look under your seats, and you should find a package of seeds. Everybody got them? Okay, good. I would like to transport you from Washington, D.C. to this rice paddy, and I'd like to imagine that you are this farmer.

The packet of seeds that you have, from these seeds you have to grow enough of this crop to feed your family, and you have to sell at the market. But here's the catch. You have limited irrigation, you have a feature phone, and you really have not that many resources. So what do you do? The short answer is that you would likely have generations of knowledge from your community on how to actually grow this crop.

[00:26:00]

What happens when generations of knowledge falter? What you'll see here is the historical 100-year average of rainfall in this specific region for you as a farmer. But what I really want you to focus on is the fact that the timing and the quantities of this rainfall is changing rapidly, and that can be incredibly hard for farmers to predict. My name's Terrina. And I'm Shweta. And we're from the Development Innovation Lab. And our lab was founded by Nobel laureate Michael Kremer. We work very closely with partners at Precision Development, and we're going to talk about AI-based weather forecasting for smallholder farmers.

So, the reality is that farmers make a huge number of decisions with massive uncertainty. There are 418 million smallholder farmers in the world, and our research has found that misjudging rain can have catastrophic effects on their lives and their livelihoods. And so, what we're doing is trying to move from their pure instinct to localized and specialized information and intelligence. And AI can enable the generation and scaling of new forecasts for farmers at record-breaking times. Our goal is to reach 100 million farmers by 2028.

We started with 10% of that goal two years ago, and we're already more than halfway through that. And here's how. So, think about the arrival of the monsoon rains. Scientifically, that's a complex phenomenon. But for a farmer, say you are the farmer today, it's one thing. It's one question. When can I plant? And that translates to something specific. Enough rains, no dry spell, and soil wet enough to sow. AI can enable processing multiple models with years of weather data in minutes to produce the probabilities for when those three conditions will arrive at the same time.

[00:28:00]

That's a direct, decision-oriented guideline that we can present to farmers and has the farmers at the heart of it. So over two years, we have not only sent these forecasts to more farmers, but we have also significantly upgraded them in how localized they are. And that's progress that would have taken many more years with traditional methods. Our teams support governments in message design, designing dissemination protocols, and embedding feedback loops through A-B testing and monitoring surveys with farmers. One key thing that we've learned is that the forecasts matter when farmers can understand them, trust them, and act on them.

So we obsess over three things. Language, so they're designed in a way that the farmer can comprehend. Medium, so they reach the farmer through a channel that they use and trust. And now we're building on richness to go from a simple alert notification to a personalized conversational advice mechanism. But translating probabilistic forecasts to advice is quite hard. You know, 70% chance of the arrival of the monsoon rains is not a no, is not a yes. And the right decision for the farmer depends on their circumstances and their preferences.

So the LLM that we're building takes this very seriously and carefully tries to understand the farmer's needs and provides advice based on agronomic knowledge and the forecast that has been generated. Our prototype Monsoon Friend has been tested with agronomists and farmers in India, still has a long way to go, and we're very excited to see its growth trajectory in the months to come. Our goal is not to build a chatbot product. It's to build an open source package that includes prompts, data, and guidelines that can be embedded in any chatbot anywhere.

[00:30:00]

The goal is simple, more forecasts to more farmers in more regions. We want to create a world where people have AI-enabled tools to help farmers increase their incomes and also increase their resilience. If you want to find out more, come talk to us afterwards. Thank you. (audience applauds)

Thanks for the round of applause and for going first. We wanted to start with a crowd pleaser with agronomic decision trees and it's only gonna go up from there. So, Per Scholas is up next. (audience applauds)

Hi everybody, I'm Caitlyn Brazill, president at Per Scholas. This is Eduardo Hernandez who leads our AI innovation and today is going to be handling the complex tasks of advancing the slides. I'd like to start by introducing you to Janay Smith-Ramsey. Janay had to leave school at 16 when she became a mom. She worked multiple jobs and just couldn't get ahead. Janay never lacked for ambition, but she lacked opportunity. Unfortunately, Janay's story is not unique. Millions of Americans are working hard every day and cannot cover their basic needs.

While we know that employers across the US have evolving skills needs and are looking for skilled talent for them to grow, Per Scholas for the last 30 years has been a bridge between those employers and that motivated, capable, skilled talent through our tuition-free immersive training model. Over that time, we've trained 35,000 Americans. They've increased their incomes by over $2.5 billion. But in the last year alone, 75,000 people raised their hands and said,

[00:32:00]

I want to join Per Scholas. And we had seats for 5,500 of them. And so we ask ourselves the question, how can we scale that opportunity? How do we reach the thousands of Americans that we couldn't give a seat to? The answer is not just building training capacity. The answer has to be scaling personalized instruction that aligns to true employer demand and connection to real localized job opportunities. And so the answer that we came up with was rethinking our workforce training system as a whole.

Enter MobilityOS. This is our integrated AI-powered learning management system. MobilityOS will be an end-to-end ecosystem that unifies AI-driven tools along with human coaching at each stage of our learner journey, from application through technical skilling, through professional development, job attainment, and lifelong upskilling. MobilityOS will also enable us to turn employer insight into real-time changes and updates to our curriculum, ensuring that we can stay aligned with evolving market demands. We can envision this system because already today we have deployed and built through this cohort and partnership an AI-native classroom that our learners are receiving personalized instruction, personalized labs, and an AI-empowered tutor who is available to them 24 hours a day.

[00:34:00]

ensuring that people like Enrique are building skills on his own time, outside of when our instructors are available. Additionally, our instructors themselves have transformed their knowledge and insights about our learners' experience. They can identify challenges that learners are having, specific topics that they're struggling with, and prepare remediation plans on the spot to help our learners through those challenges. The impacts that we are seeing are incredible. Our early AI tutor actually already found a 25% increase in certification rates. We've seen about 700 hours of one-on- one instruction that our learners are taking advantage of while in their cohort, and much of that time is happening in evenings and weekends when our instructors and staff wouldn't be available.

Altogether, this is creating for us a clear path forward to turn this model into a scalable workforce solution, taking the evidence-based Per Scholas model that today we directly implement in 25 cities across the U.S. and making it accessible for partners to deploy, expanding opportunity to that market-aligned training and career opportunities at scale. We'd love for you to consider joining us in this pathway. A $3 million investment will enable us to take that core platform and expand it across Per Scholas, touching tens of thousands of learners a year.

But a $7.5 million investment will allow us to fully deploy MobilityOS to the ecosystem, creating opportunities for as many as 30,000 people

[00:36:00]

annually to see that kind of economic mobility. That's 30,000 people like Janay, who today is a data center technician at Google. She left Per Scholas and started earning three times what she was making before her class. And today, her children have that financial stability that she was always searching for. We hope that you will consider making that investment to help us create this kind of generational change for thousands more Americans. Thank you. (audience applauding)

I swear he's one of the most technical members of the cohort, much better than just slides. Just a quick glimpse into the future on that. That number that stood out to me as 75,000 applicants and 5,500 seats. Imagine the feeling of wanting and desiring to better yourself and having to have that kind of limitation stop you from doing it and how possible it is to have every person that raises their hand, you can say yes to them. So up next is Community Economic Defense Project.

Janet and Alejandro built a stable life in Denver with their three kids. Alejandro was a warehouse manager and Janet managed an office. A medical emergency wiped out their savings and Janet missed two months of work. They fell behind on their rent and their landlord filed for eviction. They had defenses that could have stopped their case. But with a six week wait for legal services, they didn't understand the process, they missed their court date, and they automatically were evicted by

[00:38:00]

Default without ever seeing the inside of a courtroom. My name is Lois Lupica. And I'm Paul Gilson. And together with our team, we built MyReply. MyReply is not a generic chat bot. It is an AI enhanced eviction defense tool designed by technologists and lawyers who have represented hundreds of renters in eviction cases. It is grounded in years of empirical research about what really helps self-represented parties win their cases. Every year, 3.6 million American families are evicted. That is 10,000 a day, belongings to the curb, nowhere to go.

And here's why. Only one in a hundred have a lawyer. And nine out of 10 never even respond to their eviction notice. Most people get evicted by default. They lose without ever going into the courthouse. And what that means is that, oh, what happened? Sorry about that. It's not because they lack rights or defenses. It's because they lack legal help. And an eviction cost a family $15,000 in lost wages, emergency care, and moving expenses. And that is a debt that can follow a family for many years.

And that's shameful. So we decided to do something about it and we built MyReply. MyReply helps

[00:40:00]

renters navigate the eviction process, assert their legal rights, and buy critical time for people to stay stably housed. With MyReply, renters can take photos of their legal documents, and the critical information is uploaded. MyReply then identifies challenges to the eviction from a database of over a hundred defenses. And it combines these defenses with user inputs, as well as transcripts from guided attorney interviews, to generate legal advice and guidance in simple and empathetic language, all the while protecting the renter's privacy and confidentiality.

At the end of the process, the renter gets a ready-to-file eviction answer, guidance to help them get to the next step, and a plan to negotiate with their landlord. And we have just started to test MyReply with users. In early testing, we can see that MyReply is working. Some of the initial user feedback we've gotten suggests MyReply gives them a sense of agency and confidence to know what their next step is. We are rolling out MyReply across Colorado initially to serve 1,000 renters by early 2027 in our CEDP courthouse clinics, and our plan is to quickly grow from there to over 10,000 new users per year.

We know that people all over the state could use this kind of support, and fortunately, we don't have to reinvent the wheel. With our agnostic architecture, it's straightforward for us to redesign and adapt MyReply to new jurisdictions. CEDP is requesting $1.4 million a year for four years in funding to

[00:42:00]

I'll cover four things. Our development team, measurement and evaluation, attorney and user testing, and expansion in new geographies. The math is straightforward. Preventing one eviction saves the family over $15,000. Serve 45,000 renters over four years, and that amounts to over $600 million in community wealth preserved. On that $1.4 million annual investment for four years, we can deliver $124 in social impact per $1 invested using GitLab's ROI methodology. Great. Ultimately, our goal is to put my reply in the hands of every renter facing eviction in this country because we believe that access to justice should not be a privilege for the 1% who get a lawyer.

It should be a right for everyone. Scan this QR code to learn more. Thank you very much. Thank you so much. I'm going to let that hang up there for just a minute. I'm going to ask for a couple of phones to be up in the air to scan that code, learn a little bit more. In today's technical object lesson, an incredible project that will serve millions of people being thwarted by a $10 Bluetooth dongle. Such a good object lesson for how these projects work and the kind of work we have in front of us.

Okay, we are up next with MIT Media Lab. Hi, I'm Ayush from Project Iceberg. Meet Maya. She's 17, lives in Westlake, Ohio, and is about to make one of the most consequential decisions of her life. What should she study? What skills should she acquire? And what kind of work will actually be there for you when she's ready to graduate? Now, this one choice she's making now could potentially shape the next 15,000 days of her life. And she isn't doing this alone.

She has her parents, her school council

[00:44:00]

the colleges she is applying to, and even the state of Ohio there to help her. However, all of these people rely on the same signal. It's job posting data, employment surveys, statistics about wage or enrollment data. Now this data explains the economy of today, but Maya, she isn't preparing for that. She's preparing for the economy of 2030. And the signals that we rely to steer our economy today explain less than 5% of the world that Maya will step into. And this is just the tip of the iceberg, because Maya's decision doesn't just affect her, it cascades across the entire economy.

It reshapes what colleges will teach, the kind of talent employers can hire, and eventually reorganizes the entire communities around it, from where people live to who the Starbucks on Main Street serves. So we built Project Iceberg to see this cascade before it shows up in data. Our platform consumes thousands of data streams about how humans do work today, and how AI is actually reshaping that work at the level of tasks and skills. We quantify this exposure with the Iceberg Index, which can be measured for each zip code and each worker across the entire economy.

And then our proprietary simulation platform allows us to simulate how this change cascades through millions of people and institutions across the entire economy. So here is a quick video of the platform in action. Now Maya's high school counselor can use the Iceberg platform to see how the entire labor market is evolving at the level of states. She can drill down to see how these patterns vary across different counties within the state, going down into the county where Maya comes from, and observe exposure patterns down to individual neighborhoods, worker personas, and people across the economy.

Now imagine Maya wants to work as a financial analyst. The platform doesn't just tell us

[00:46:00]

how the occupation is changing, but what this presents as an opportunity for her future. Can she find work where she is living right now? Simulations help us see how employment may change and where she may find these jobs. Now, Iceberg is built through years of research in our lab at MIT. This is basically the outcome of my PhD research, where we invented an entirely new class of AI systems called large population models. Instead of using one AI to make predictions, LPMs allow us to simulate entire populations with millions of people, capturing how individuals respond to decisions and how risks cascade through their interactions.

So we used Iceberg to deploy a digital US on Frontier, which is one of the US's largest supercomputer down in Tennessee, and this platform already helps us provide foresight into many critical challenges across the economy. And it's already protecting all of us. Iceberg here is helping us bring life-saving vaccines to your doorstep and also securing food supply chains so that egg prices stay in check. Now, with support of the GitLab Foundation, we focused the last six months on turning Iceberg into a foresight platform for the new labor economy.

Today, Iceberg monitors over $1.4 trillion of exposure to how work is changing across the US economy, and institutions are already using it to act in multiple ways. They're helping students like Maya figure out what they should learn. Employers are using it to figure out how they can reskill their workers. And state governments, senators, and even presidents across the world are using it to figure out how they can reshape their economies and infrastructure to deal with this new reality. And it's been featured extensively in the press.

And for institutions that use Iceberg, it's becoming a decision co-pilot that helps them simulate scenarios, stress test interventions, and optimize the cascade before they come into billions of people

[00:48:00]

salaries to implementation. Now, there are millions of Mayas around the world, just like Maya in Westlake, Ohio, who are trying to make decisions, figure out where their next opportunity lies. Our goal is to transform Iceberg into public decision infrastructure that foundations, policy makers, and enterprises around the world can use to help navigate the new economy and act before change arrives. So we want to work with you to bring foresight to the decisions that you already find, helping guide students, empower workers, equip employers, or even prepare communities for the new reality.

So with Iceberg, we want to bring foresight to the new economy by helping you simulate your decisions before you deploy them in practice. Thank you. Well done. I get that foundation. We often talk about AI has really forced many institutions to move from long-term planning to navigation. I love this concept of public decision infrastructure and scenario planning that Ayush is working on. Okay, off to Moses/Weitzman. Thank you very much. I'm Cassandra Okechukwu, and I'm the Chief Science Officer at the National Association of Community Health Centers.

Previously, I was the Director of Research at the Moses/Weitzman Health System, which runs one of our health centers, our nation's health centers. They're really an untapped engine of economic mobility. Listen to me. Health centers serve 52 million patients, one in three in rural areas, and by statute, they have to be in low- income neighborhoods, making health centers a way that we can get to people who need economic mobility. People like Kisari. Kisari is a worker that I

[00:50:00]

I met at the Moses/Weitzman Health System, whom Dr. Katrina Yamazaki supervised. And she was a grocery store worker who was trained by Weitzman to become a clinical coordinator. Well, you know what happened in the last two years? There was defunding. And so I was potentially having to maybe lay off Kissairy and others who worked on this project. And I was thrilled to hear that Dr. Yamazaki had prepared them for the layoff by getting Kissairy certified in phlebotomy. However, when we had our check-in, I found that she wasn't planning to stay in healthcare.

She was planning to go back to the grocery store work because that's what she knew, jobs travel in social circles. And the thing that is vexing about this, about the plan that Kissairy has, is another problem we have in this country. The jobs above those jobs occupied by people like Kissairy are mostly empty. So we are down in this country about half a million higher wage workers, and yet people like Kissairy are moving from job to job and only making the same or less wage that they've ever made.

And so with the help of the GitLab Foundation, we are trying to solve this problem. And we're working to solve this problem by coming up with this career pathway engine. So it's an engine that understands the worker, focused on FQHCs, workers in the FQHC system. And it maps out several pathways for these workers to not just find the next job, but also for them to be able to map a career for themselves. So that's what it does. And it coaches the workers along the way using plain language.

Another important thing is that it unlocks benefits and resources. And this-

[00:52:00]

So this is my sister, Vanessa. She lost her job at an FQHC where she was a social worker while doing her cancer treatment. She stood not to only lose that job, to only lose her salary, but to also lose $30,000 annually in loan repayment that was associated with working at community health centers. And I was able to get data, work with her to help her find another health center where she could get a job and still maintain that $30,000 in loan forgiveness.

We shouldn't have to have a sister who's on Harvard's faculty to be able to get loan repayment that our tax dollars are paying for. And so in Kissairy's case, what we've designed is something like this first pathway, something that the engine will give her. So with this first pathway, it shows her if she stays in the FQHC system, all those loan repayments and scholarships that she can get to allow her to move from a recruiter to become a nurse and then move on.

And it does this using language and using resources specific to FQHCs. However, knowing the way the labor market works, we've also allowed that engine to give her exposure to jobs outside the FQHC ecosystem. This is Kissairy. And she just had a baby this summer. She still has her job, and we're very happy about that. And we love having her in the FQHC system. However, we recognize that data is power, and part of what we're doing is democratizing data so that if we want to keep people like Kissairy in the FQHC system, we are able to show them the pathway.

But then they also have the

[00:54:00]

data to take their talents elsewhere. Because this is something that is counterintuitive, but it's actually true. Ellie can tell you about this, that when you provide pathways to economic mobility, you actually increase retention. Walmart has done it, and it's been done. And so we see this as a tool that doesn't just help the workers, but it also helps the workplaces, especially those who really want to move their workers up and use them to fill the gaps that we have. And so working with the National Association of Community Health Centers, where I am, we've come up with a plan, a two- year plan, to get this product to many health centers, to 300 FQHCs.

And we want you to join us in making this possible. We have a modular budget that we can share with you when you come to talk to us. And we want to hear from you, not just in terms of the funding, we love that, but we also want you to please come give us advice, come give us, just talk to us, because we are new to this, but we are excited and we're here to stay. Thank you very much. I love that project because it's such a win-win.

If you think about the concept of mobility within large companies, there's so much human effort and time spent on moving people up through companies into higher skilled labor, higher skilled opportunities. In a place like an FQHC, this is both providing additional support of healthcare and also creating pathways to close these talent gaps. And everyone wins in those scenarios. And it's such a great niche invention. Okay, Career Path Services is up next. This is Hameeda. Hameeda moved to the United States three years ago from Bangladesh with her youngest brother and her mother.

She's the main provider of her family and she's ready for what's next.

[00:56:00]

And healthcare is hiring. There are over 100,000 healthcare jobs in Washington state alone. This summer, Hameeda became the first user of TruePath. I am Harpreet Marwah from Career Path Services. This is Arundhati Roy, and we built TruePath to help people like Hameeda navigate what's next. This is what stands between Hameeda and all those jobs. She has to first figure out which career should I choose? What does the labor market data say? How many openings? What does the roadmap look like?

And with all this news of AI taking away jobs, which jobs will be AI resilient? What matches with her skill sets, her interests, her life's constraints? And after she figures that out, then which programs will help her get there? What community colleges? What certifications? What apprenticeships? And then what kind of funding streams will help her pay for it? And underneath that, the barriers that quietly stop people, being able to get dependent care, being able to find English language classes, and being able to pay for transportation.

Each of these resources was built to help Hameeda, but it's really hard to get to. And her case manager is excellent, but he has 45 minutes to navigate a system that takes people years to learn. And so we built the front door to this support. This is not meant to replace Christian, the case manager, but it is meant to place decades of system knowledge in his hands and hers. With a few easy language questions that anyone can answer even on their worst day, with limited digital skills or limited English skills, this product then tells them three attainable pathways that have openings in her locality and that have trainings that she can reach to.

Under the hood, we are using verified Washington workforce data. We are using lists of

[00:58:00]

eligible training providers and directories of support services and we are using AI to make it easy for Hameeda to enter this information and to understand it and to ask follow-up questions. And then Hameeda walks away with a plan with her name on it, with the job title, the training program, the wages, the timeline, and the next step. For someone that walked in that thinking that the maze was her problem, she now has a page with a path and her name on it.

Career Path Services is an organization that has been serving job seekers for over 50 years. We serve over 6,000 job seekers a year and we help people find meaningful work, build skills, and reduce barriers between them and economic stability. We are the one-stop operator for six out of 12 of Washington's workforce development areas and we have our own training program called MediWork$. We have a partnership with Seattle King County Council and that is where I also work and lead an operator team.

And that helps us expand our scale to potentially 25,000 more job seekers who they serve every year. We are live today with 20 of our own case managers who are giving us structured feedback. And just last two weeks we had around 150 job seeker sessions. But the result that really excites me the most is that before using TruePath, half of these job seekers were not really considering any healthcare opportunities and after using it, 80% are. That shows me that this tool is doing what it was meant to is open up people's or shift people's mindset towards new opportunities.

We are seeking $1 million for three things. We want to expand beyond healthcare to two to three other high growth industries. We want to improve our product and increase features beyond one-time assessment to ongoing navigation for both

[01:00:00]

job seekers as well as the frontline staff that support them and we want to build a verified data infrastructure that this space desperately needs. Hameeda did not need another program what she needed was a way in and there are thousands of others that are on the outside looking in and we need your help to be able to support them. Thank you. I love that stat. For those that work in career navigation tooling for a career exploration for students sometimes the smallest little experience in their life can direct them to a new career pathway and that that number of moving from 50 percent of even knowing there's an opportunity for a health care job to 80 percent with an extremely light touch intervention.

Just imagine the changes that that can make in people's life and in talent shortages. Up next we have the National Association of State Workforce Agencies. Hi good afternoon everyone I'm so excited to be here today and today I'm going to tell you a story that's based on a real life person that we met when we were doing our early user research. Our project team affectionately calls him turtle guy. If you were to look at turtle guys resume you would see one thing dishwasher.

In our early user research we had the opportunity to ask him more questions. What are your skills what are your interests and he said I don't know I'm just a dishwasher. I'm Allison I'm with the National Association of State Workforce Agencies and I'm Kelly I'm with the New York State Department of Labor. Through this conversation we found out that this person was really into his pet turtles. He shared that when his turtle tank broke and he couldn't afford a new one he took it apart he figured out how to fix it and now he goes to

[01:02:00]

He buys broken tanks, he repairs them, and he sells them to friends in his turtling community. Most job sites make recommendations based on your work experience. They require a job seeker to know what they are looking for. If Turtle Guy had applied for an apprenticeship or entry-level manufacturing position, the company's resume software or staff would most likely screen him out. Why? Because he doesn't have the right keywords, certifications, or experience. Just work history as a dishwasher. This candidate had the aptitude and potential, but the system couldn't see it.

Our approach is to improve that matching process, and we do so from an individual's own experience, both their personal experience and their work experience. We are focused, for now, specifically on registered apprentices. But let me pause for a moment. What comes to mind when I say the word apprenticeship? Is it something like our medieval Middle Ages baker up here? Is it Wart from Sword in the Stone, who's apprenticed to Merlin? Well, apprentices have changed. They are everywhere. Today, they offer a host of opportunities for someone to earn while they learn.

And a lot of them provide a national credential upon completion, creating that kind of economic mobility that we're looking for. Today's apprentices, they work as electricians, locksmiths, they work in semiconductor manufacturing. But even though those opportunities are there, it's not always easy or straightforward to find them. New York State alone has more than 900 programs and currently has about 18,000 active apprentices.

[01:04:00]

One of the toughest barriers is awareness, for someone like Turtle Guy to even know these opportunities exist and to recognize that they have the potential to succeed. Our registered apprenticeship matching tool, talk with Tyler, helps connect people with great earn-while-you-learn options. It doesn't ask for a resume. Instead, through a chat feature powered by Generative AI, Tyler asks about interests and experiences. It prompts for what people like and dislike and encourages reflection. In just a few minutes of conversation, Tyler can get beyond just a dishwasher and identify a problem solver with electrical aptitude.

With these insights, our tool provides matches to New York State registered apprenticeship programs that align with the user's interests, not just their work history. These jobs pay from day one and have strong earning potential. And we know that there are thousands of people like Turtle Guy across the country who are looking to positively impact their future. Our tool can help uncover that hidden potential, connect people to registered apprenticeships in these high-demand careers. They possess the drive, the talent, the skills that businesses are looking for.

But they need that additional guidance, support, confidence to connect with those opportunities. And there are more of these opportunities coming online every day. There are 26,000 active registered apprenticeship programs nationwide, and more than 3,000 of them have been created in the last 18 months alone. We're seeing this resurgence in the interest in the apprenticeship model, and these are jobs that are often the least affected by the AI revolution. HVAC technicians, plumbers, steam fitters, even my favorite, brewers and distillers. If your organization is passionate...

[01:06:00]

about matching people to opportunities that accounts for their range of skills and expertise, come talk to us. We have candy. We have a lot of bridges to build, and we would love for you to join us on that journey. Thank you so much. So Paige, maybe you could start. Can you tell us a little bit about yourself and your organization and what you guys are up to? Yeah, for sure. So I'm Paige. I'm the co-founder and executive director of the Student Basic Needs Coalition.

And we're working to support the three in five students who experience basic needs insecurity at some point during their education. When I was in college, I worked three jobs. And even then, I sometimes skip meals to make ends meet. And I graduated, but millions of students who experience those same issues don't get that chance. So I started the Student Basic Needs Coalition to solve this problem. And now we build technology that makes it easier for students to access the support and easier for institutions to deliver at scale.

Our goal is for administrative complexity to not be a barrier in the way of students getting a credential, a degree, and achieving that long-term economic mobility. Thank you. And Dustin, what about you? Can you tell us a little bit about GiveDirectly and what you guys are tackling? Sure. So hi, everyone. It's great to be here today. I work at GiveDirectly. We're a global nonprofit. We deliver cash transfers to people in poverty. We always say cash is one of the most studied anti-poverty interventions in the world.

We've been doing it for a long time. And we also do it here in the United States. So a lot of the work you might know is for our international work. But we work here supporting people across the country in different life circumstances with cash at kind of the right moment to get back on their feet or achieve their goals or anything like that. And Dustin, sticking with you for a second, can you just tell us one data point that you guys are particularly proud of?

Yes. So one program I'm really proud of in the United States is called RX Kids. It was a focus of our participation in the cohort last year. And so we provide a cash transfer to a mother who's expecting in that

[01:08:00]

prenatal period, and then cash in the first six months or 12 months in some cases of the baby's life. And this is flexible, unconditional cash. But the stat that I think is just amazing is what we found in the city where we started Flint, which has Flint, Michigan, which has the highest child poverty rate in the country for a city of its size. After receiving this transfer, the neonatal intensive care unit admissions, the NICU admissions, dropped by 29%. So if you work in public health, we say this is a health program, not a cash program.

It's a cash prescription. And the health outcomes we're seeing are things that public health departments would move mountains to have happen. And what we're really seeing is moms trusted with cash are finding ways to deliver much healthier babies, which I think is just incredible. So 29% drop in NICU admissions. That's incredible. I feel like I spent a lot of time reading economics papers in college, and we used to get really excited when you would see just a couple of percent drop.

So I mean, 29%, mind-blowing. Thanks for sharing that. And Paige, same question for you. Can you tell us about a data point you guys are proud of? Yeah. So over the past 18 months, we've helped refer over 60,000 students to over $150 million in basic needs resources. And I think that this really shows the scale of the problem, the demand for the solution, and the opportunity in front of us. But the next step here is really being able to determine what resources students are actually receiving, how long they're staying on them, and what the long-term outcomes are for their life.

So that's what we're really focused on over the next couple of years is building out that measurement infrastructure to see how these solutions are shaping their lives going forward and contributing to that economic mobility. Incredible. Thank you. So both of you are leading really phenomenal organizations doing high-stakes work, but you're trying to keep your teams lean. Can you talk a little bit about what the day-to-day implementation of AI

[01:10:00]

look like in your organization and maybe a few of the things that have surprised you as you've rolled it out? Yeah, so I think that AI is at its best when it helps simplify the repetitive processes that human judgment isn't needed for. So in our work, that looks like when campus staff would usually be answering a bunch of questions over and over, or when students have to tell their stories repeatedly to get access to the support. Instead of doing that, right, they can work with a technology platform that can automate these repetitive processes and really give people that time back to focus on the things that need their judgment and can contribute to that one-on-one connection and sense of belonging.

And I think within that, one of the things that's really surprised me is how AI has quickly become really good at delivering information. But the next step there that we need to think about is how to turn that into action. So making sure we're getting to that next step and helping people navigate these really complex processes in a way that advances their long-term goals. Yeah, and I think from our side, I'm equally interested in the staff kind of empowerment, the internal operations.

There's a saying in the civic tech world that the strategy is delivery, like doing the thing is where you learn and execute. And that's why anyone should trust us with their money to get it to somebody, frankly. And there's so much happening internally that I'm excited about of kind of smoother flows, more efficiencies. A big part of our brand at GiveDirectly is getting the maximum amount of dollars to recipients, that really lean, efficient operation. We kind of obsess over that. And so ways to make those flows better are super helpful.

And then on the recipient side or your client or customer, any term you want to use, moms and babies for me, the tool we built as part of the cohort last year, moms have to submit a document, which is already a theme in some of the things you heard today. And so what we kind of landed on and iterated was when a mom submits that document demonstrating pregnancy in real time, it's telling her if

[01:12:00]

That's the right document. If it looks similar to things that we need to assess eligibility, if it's blurry, if it's unreadable, something like that. And so there's a lot we can do to minimize burdens, but there's still things where it rests on the recipient's shoulders. And that kind of simple, not that sexy AI intervention actually really helps moms get what they need in that first go. So we're not going back and forth asking for more documents. Again, public health intervention, we want them to get the money as quickly as possible.

You can imagine how babies are growing. So anything we can do to speed that up is really helpful. That's where we've seen some AI gains. That really resonates. Get the money out as quickly as possible. I like the strategy. The strategy is delivery. That's a great trope. All right. How have you been navigating moving fast with AI tooling while still protecting the trust of the communities you serve? And maybe I'll go for you first. Sure. Yeah. So we have an amazing technology team internally that I think has done a lot to be thoughtful about standards and clarity and communication internally at the organization.

You are responsible for the output of the tool you're using. The human is there. There's not kind of sidestepping that. I see that as a principle in a lot of projects. I think that's super important. One small example is the tools we found, again, in the global work are really helpful in translation, like even better than you'd expect. But there are languages that are underserved. So we have a smaller project. We run call centers. We have a lot of data of interacting with recipients, often subsistence farmers in various African countries.

We have a program running right now to fine tool the language access for models like this to make sure that it's reaching everyone. How about you, Paige? I think balancing speed and trust in this field is so important. If you send your AI agent to send out an email to someone and it does something wrong, that's annoying. But when you're working in a field like this and you give someone the wrong information about resources that are available to them or

[01:14:00]

know how to access them, that's less money in their pocket. So because the stakes are so high, our intentions and our outcomes and the quality have to be even higher. And for us, one of the ways that we're really investing in this is by co-designing with students and with administrators on these college campuses. We've done a lot of student leadership work over the past couple of years. We're really diving into students' experiences where they're seeing the barriers and their ideas for solutions too.

You know, not just asking people about the problems that they're facing, but really getting their perspective on what the right solutions are. And so I think that that's really critical as we think about building this space. All right, for you, Dustin. GiveDirectly has distributed over $1 billion in cash directly to people worldwide in poverty. As AI reshapes labor markets, does the case for direct cash transfers get stronger or more complicated? Ellie gave me a preview of this question. I'm still struggling to answer it, so bear with me.

I think it's super complicated. We're a research-driven organization. We're founded by economists. We've had over 30 randomized control trials of our work. This is the type of thing we like to engage on. And I've been on my own kind of learning journey in the space for the U.S. labor market disruption coming, and I think a lot of folks don't know what to expect. Is it young workers, new workers, white-collar? How soon are we going to be in an economy that looks really different or kind of just accelerated?

And I don't know if we know those answers. I guess the two models for cash, it depends on what you think of it as doing. If it's a bridge when something goes wrong, and we think there's going to be a lot of cycling in the economy and people are going to be switching jobs more, I could see a lot of upside there. Just as an example, in the fall, when the SNAP program, the food assistance program that folks work on here paused, we ran a really rapid program with some partners.

[01:16:00]

tell and others to deliver like a bridge payment, because we knew even just a week or two of disruption can really throw someone into a financial spiral. So we paid almost a quarter million people in just a couple of weeks then, and hearing from recipients, that put food on the table. The other model in our international work is what we see when we give a lot of money to somebody up front, really, really poor people. So $1,000, but that could be a year's worth of resources.

They're super entrepreneurial. And so that's kind of a launchpad to do something totally different, to get to a different rung in the labor market, to start a business, to invest in assets that are income generating. And so I'm curious if we will see that pattern in the U.S. as things change, and what kind of levels and modalities of delivering the cash matter the most. I'm guessing what our system is now that was set up a long time ago probably could use some iteration and testing, something I'm thinking a lot about.

But I think the jury's still out on that. That's fair. I think you may need Project Iceberg to help you with some of this forecasting. All right. For you, Paige, your work connects students to public benefits through technology. Benefits access is very in vogue right now. What's the next step to ensure scale, and where do you see it moving? That's a great question. So I think as a field, we all know that we've gotten really good at telling people what they're eligible for.

But the next step is actually making sure people receive those resources and stay on them and achieve the long-term outcomes. And the challenging part about that, right, is that the data lives in government offices and institutions that we as nonprofits or funders don't always have access to. But it's really critical that we get that information because as we're able to better understand the data behind what we're doing, we'll also be able to experiment and influence the interventions that will really drive the outcomes that we care about.

So the next phase, the next thing to scale this industry is building

[01:18:00]

that bridge between the data and the impact. That makes sense. I think this is for both of you. We talk about AI as a tool for economic mobility, but tools only work if people can access them. What's the access gap that worries you most? I'll start with you first, Paige. Yeah, so from my perspective, AI has a huge opportunity, has huge potential to either widen opportunity or widen inequality. And I think a lot of the conversations that I hear are very bimodal about AI.

And they're very focused on like, okay, this is like the end of the world or like, this is going to save everyone. And, you know, I don't know if I really agree with either of those. I think that AI and any technology is only as good as the systems that it lives within and as the biases and experiences of the people who are building it. So, you know, as we're thinking about this next phase of technology, it's really important to think about who we're building with, thinking about the systems that are in place and making sure that we're not just like replicating existing bad systems and we're actually creating new ones.

I think that, you know, the biggest thing that I want to think about is how we can make sure that AI is not just making existing inequalities faster and more efficient. Yeah, that's a great answer, Paige. When I hear access gaps, I just think of our safety net. That's kind of what I've worked on the most. And right now there's this long tail of changes that's kind of building momentum where lots of people are losing safety net benefits. In Arizona, based on the best data, about half the people have lost their SNAP benefits in about the last year.

And some of that's from new eligibility, but a lot of it is just from a failure to navigate administrative burdens and kind of reporting and things like that. And that's kind of heartbreaking when we think about all these new tooling and potential for making it really easy to hold it together and kind of demonstrate you are who the state needs to know you are and you're eligible and things like

[01:20:00]

So I think that's really top of mind. We're trying to think through, building on our work last year, creative ways to kind of meet that moment. But one quick positive example, at my last job, I worked a lot with state frontline staff at state delivering benefits. And our engineering teams would always say the best hackers in the world are these like 20 year veterans of, you know, caseworkers in Medicaid or SNAP or whatever. They're using these systems that we, you know, as a taxpayer, as Americans, spend hundreds of million dollars to build that are so hard to work with.

And so I would love these tools to get in the hands of those frontline experts because I think that democratization of access would be pretty incredible. I think it would really help people do awesome work. And get those old school hackers to figure it out. All right, last question on my side. If AI delivers on its promise for economic mobility, what does the world look like in 10 years? I'll start one once more with you, Paige. I think my vision for the next 10 years is that complexity stops being a barrier to opportunity.

So right now, as people are navigating all the, you know, transitions that are coming into play, especially with this new AI economy, right, they're thinking about applying to jobs. They're thinking about figuring out childcare so they can go to school. They're navigating financial aid applications. And all of those, you know, points add friction. And friction takes attention away from the things that matter most. So I think that when technology is at its best and when it's being used effectively to solve these problems, people will be able to give their attention back to the things that really matter.

You know, students can focus on learning, on building their futures. Staff can focus on building that one-on-one connection with the people that they're working with. And we'll all have more space back to focus on building a better future. But it is special. It's pretty rare in the wider funder landscape. Sometimes there's onus requirements or it's sort of you're on your own, come back. And this is like a great blend around that. I think that's really special. The other thing I was trying to think a little deeper about this question,

[01:22:00]

I think there is a lot of support for technology in nonprofits. There's a lot of interest in there and funding. The operations side is probably under-invested. For a delivery organization like us, that's so critical. We're grappling with problems of how can we best procure hundreds of thousands of phones for our recipients across multiple countries at a good price that we trust in. Kind of just like operations experts, I would imagine inside of some of these leading newer companies, there's so many cool things happening that we could learn from, from kind of a management delivery perspective.

I know that would be really welcome by us and probably everyone in the room. That's very helpful. Thank you, guys. We'll take that away and think about it. Thank you both. Yeah, thanks so much. Bear with us. We have a couple more great ones to go through. We're going to go to the National Domestic Workers Alliance. My name is Ai-jen Poo, and I'm the president of the National Domestic Workers Alliance. So great to be here. We represent the two and a half million nannies, cleaners, and home care workers who work inside of our homes supporting our families.

And we're so excited to share with you our AI assistant for domestic workers that we built. And our organization lost a dear friend yesterday, Gloria Steinem, and we wanted to dedicate this presentation to her today. Hi, my name is Elza, and I've been a home care worker for almost 30 years. I've helped elderly clients heal from serious illness and made sure that they felt safe and cared for. But most of us do this work alone. We have no co-workers. We have no HR.

We have no one to call when a client doesn't want to pay us or we've been badly treated or unfairly. That's why

[01:24:00]

Ask Aya matters. I know a worker who asked Aya if her pay was fair. She hadn't received a raise in many years. Ask Aya showed her what workers in her area were making and helped her practice asking for more. And guess what? She got the raise! For a worker, that means putting food on the table. You may know tech, but I know care work. I know what it's like to be afraid to ask for a raise or a day off, to fear retaliation just for speaking up.

I've never imagined that I would be designing technology, let alone AI. But NDWA built this with us in the center, every prototype, every principle was shaped by domestic workers just like me. And the need for care on the part of American families is surging, especially for a growing aging population in our country. It's the reason why home care is always among the fastest growing occupations in our entire workforce. It's also among the most undervalued and insecure work in our entire economy today.

Meanwhile, AI is transforming the world of work in profound ways. The question is not whether AI will change jobs, it's really how soon and for whom. And right now, AI is happening to workers. Jobs are changing, we're being managed by algorithms, some jobs are being displaced. What if AI

[01:26:00]

AI could be built by workers for workers. And what if AI could actually help make precarious jobs better? The time is now. In the next 18 months, we will know whether they will, the next 18 months will define who benefits from AI and we have to make sure that the workers who are on the front lines of its impact have a say. Enter Ask Aya, our free bilingual AI assistant for domestic workers that was built by domestic workers for domestic workers.

Workers were involved in every step of the process. Workers defined how data would be stored and how privacy would be protected. Even the name and the color and made sure that the logo though that Aya had her hair up in a bun so she looks like she's ready to get to work. It's really truly amazing and our beta results speak for themselves. 91% net promoter score or 91 net promoter score, 93% took real action using Ask Aya's advice and 25%, one in four workers, were able to negotiate a pay increase in six weeks.

Well, Ask Aya helps workers navigate the moments that matter most from asking for a raise to a sick day and really, really importantly, Aya connects workers to local organizations so that they can find community too. When care work is good work, the entire economy is stronger. It provides security for millions of people and all kinds

[01:28:00]

and what we built could be applied to most low-wage jobs, gig workers, restaurant workers, farm workers. I want to thank the support of some of you in this room. We already have a real impact on the lives of workers, but now, today, we are ready to scale up. All of you are in rooms where decisions are being made about how AI will be developed and who will be at the table, and we need advocates in those rooms, advocates for low-wage workers like care workers, and we also need partners to scale.

For six million dollars over the next two years, we believe that we can create and build Ask Aya into a world-class standard bearer for dignity-driven AI, for domestic workers, and beyond. Thank you. What a wonderful project. We are going to be next with Colorado Thrives. So I'm often on the floor with my three young kids playing board games, and there is one board game that is a fitting metaphor for the labor market today. Shoots and Ladders, you may be familiar. In our labor market, folks are trying to make it, right?

We're trying to have living wage, stable jobs that provide a pathway up, and as we know, especially with AI and our economic cycle disrupting what we know of typical jobs and pathways, there are shoots and there are ladders. The ladders are those evidence-based programs that give great results and get you to that living wage, those academic programs that you know will lean you into a good job, and then there are shoots that are dead ends or that keep you in that hourly job where you don't have

[01:30:00]

Upward mobility. In Colorado, what we know is six in 10 Coloradans who get a credential in our state are not making a living wage within five years. There are too many shoots in Colorado and not enough ladders. And our driving question is where are the ladders and how do we know? I'm Christine Heitz, and this is Jesse Olson from our partner at Elevate Quantum. And we're here with a demand side and an ecosystem view of how do we solve this. I lead an organization called Colorado Thrives, which is an employer, CEO-driven organization focused on economic mobility.

And we're looking at how do we bring the demand signal, demand side view, and a cross-sector ecosystem view to this problem. Our answer was to partner with our friends at FutureFit AI and build an ecosystem infrastructure that could help us learn from great data and unite a number of players in one place. This platform that we built is a job platform that really acts as a public square for pathways that we know have supply and demand. It does AI skill matching for job seekers to kind of deconstruct their skills to find opportunity.

It highlights high demand pathways where we know there's employers and real jobs on the back end, so no false promises. And I think most importantly, it serves as actionable outcome data so we now could know outside in who's getting the jobs from what programs, and we're learning quickly, and we're asking a different set of questions to solve the problems. So we are thinking about how do you unite multiple sectors and industry verticals on one platform versus duplication. So we are serving on CareerFit Quantum, as you just heard, technicians, you know, short-term credentials.

We're serving two- and four-year college students who are looking for business and tech jobs. We're serving transitioning service members who might find clear jobs or security clearance, and we're serving justice-impacted Coloradans looking for fair chance hiring and companies that really serve that population. So we're hopefully building the infrastructure and tools that can be applicable.

[01:32:00]

We think this is replicable. We're looking to scale the outcome tracking so we're scraping outside in data to see who got a job, what companies they were placed in, and what wages, and how do we use that to learn We're onboarding industry verticals every month. So we're bringing new partners on. And we think this is a replica model across the country. We're trying to get this right in Colorado. And we think it's a model that as folks are building workforce pathways, as you all are supporting nonprofits doing this work, how do we make the insights and the platform a bit easier?

So thanks. We look forward to talking more. All right, next up is SkillUp Coalition. Steve. My wife thinks I'm a coward. Seriously. As an urban dweller, the woods scare the shit out of me. Yet, this summer, I spent some time amongst the trees and I have to say I rocked it. No fear. But the woods are scary. It's easy to get lost. Hard to navigate. So I'm Steve Lee, woods conqueror and less important, CEO of SkillUp, a career navigation platform helping underserved populations.

All right, so meet Lynette. Single mom, just lost her job, can't pay her rent. What I want to see is how coach addresses the sequential steps of her journey. Come on. First, extract her skills, some skills she didn't even know she had. It then recommends

[01:34:00]

Three careers that match those skills. Not 300, three, to minimize choice paralysis and gives her the agency to pursue further. So she pursues further. She probably needs training. But Coach remembers life circumstances, single mom, financial constraints. So it recommends a low-cost online program to meet her schedule and her budget. But there's more. Coach remembers that she lost her job. So it recommends a job that matches her skill sets to bridge the gap while she pursues training. And there's more. It also positions her skills to best framework for that job.

Okay, all that is really cool. Come on, come on dude. Sorry guys. Doesn't seem to be working. Maybe I'll just do this. Okay. Okay. Where was I? Okay. Life happens, right? Lynette has to leave because life happens, right? So Coach understands, right, that life happens. And she says, I need to leave and I need to come back in a few weeks. Coach says, we got you. We're here for you when you get back. Awesome. So she comes back in a few weeks and Coach, and she starts right where she left off.

The coach remembers everything, right? He remembers all of her skills.

[01:36:00]

I think the slides are messed up. That's okay, I'm just gonna speak. Screw the slides. So remembers are her skills, her preferences, her trainings and her options that allows you to continue that journey, right? Because the journey determines the destination, right? But there's more. Lynette left not because of work, but she couldn't pay the rent. Coach understands that. So it connects her and refers her to a local two-on-one for housing support, life circumstance. It then intuits that she actually might need a human coach.

So it connects her to a real human coach through a partner of ours, all through the living dashboard. Awesome. All right. So we are a four-year-old organization. We've helped roughly 300,000 folks get out of the woods. Awesome. $18 billion in lifetime earnings. But we have a lot left to do, right? To reach our 2030 goals, we have two thirds left to do in only one third of the time. And we believe coach can be the repellent to get us there. So we're seeking 20 million over four years to help lubricate that effort.

Searching for a career can be a scary proposition. So help us help others get out of the woods. Thank you. Thank you. Thank you. Moms First is up next. And if you need me to advance your slides, I'll do that. I'm Deborah Singer. I'm the chief marketing officer at Moms First. And I'm the mom of three girls. Eight years ago, when I was pregnant with my first daughter, a routine doctor's visit revealed a life-threatening complication.

[01:38:00]

I delivered my daughter at 34 weeks, and she spent the first week of her life in the NICU. And from her bedside, when I should have been caring for her or healing myself, I had to apply for my state's paid leave program. Now, I'm like the vast majority of mothers in this country who work to support our families, and the US guarantees none of us paid leave, which is why so many of us go back to work just two weeks after giving birth.

Yet I was lucky. I lived in New York, one of 14 states plus DC, that does offer some paid leave benefits. And the reason they do is that childbirth is one of the, I'm going to get this right, impoverishing events in a person's life. Just as your expenses are ballooning, your income drops, but paid leave catches you and keeps families out of poverty. And decades of research has shown that it increases mothers' attachment to the workforce, their lifetime earning, and a family's economic mobility.

Yet three in five eligible parents don't access their state's paid leave benefits, even though they've already paid into them. And the reason, as I learned firsthand in the NICU, is that they are incredibly hard to access. You have to wade through pages and pages of a government website or wait on hold for an understaffed hotline. The result is what we call the family benefits gap. Tens of millions of parents in this country are leaving tens of thousands of dollars per parent on the table because they don't know about it or

[01:40:00]

It's too hard to access. That's why Moms First built PaidLeave.AI. Moms First is a national nonprofit organization fighting for America's moms and the support we deserve, like paid leave and affordable childcare. Four years ago, we launched PaidLeave.AI as a pilot in New York with the early support of OpenAI and Craig Newmark Philanthropies. Since then, we've expanded it nationally with the support of Pivotal Ventures, Robin Hood Foundation, and Tepper Foundation, who's here today. PaidLeave.AI does what no government website or hotline can do. It provides personalized coaching and support to parents with no extra time in more than 80 languages.

And in three short years, we've reached more than 250,000 eligible parents. To put that in perspective, that's equal to the entire population of my hometown of St. Louis, Missouri. 70% of those are parents who qualify as low to moderate income, exactly the people who need these benefits the most and are least likely to access them. PaidLeave.AI helps them understand their benefits, navigate them, and put real money in their pockets. We're talking tens of thousands of dollars, like this New York mom who used PaidLeave.AI to access $12,000 in unclaimed benefits and even discovered two extra weeks she didn't know about.

Imagine giving every person in St. Louis $12,000. That is life-changing economic mobility for families. But the opportunity head is still enormous. We estimate that parents who visit PaidLeave.AI are likely eligible for $3,000 to $8,000 in additional benefits that they're not claiming because they either don't know about them or they're too hard to

[01:42:00]

navigate benefits like child care subsidies or assistance to buy healthy food for themselves and their babies. That's why we're so thrilled to receive GitLab Foundation's support to pilot expanding PaidLeave.AI into a true family benefits navigator. Our goal is to help low to moderate income families earn the support or get the support that they've already earned, stay in the workforce, and provide for their families. In the first part of our work with GitLab, we've trained our AI model on new benefits, specifically WIC, SNAP, and CCAP, which are child care subsidies, in our pilot states of New York and New Jersey.

We are live with this pilot and have already reached 82,000 parents. We've specifically conducted usability testing with a learning cohort of low to moderate income parents to ensure that the tool is working for the parents who need it most. And now we are poised to expand nationally. We are raising a $2 million investment to expand PaidLeave.AI into a true family benefits navigator that would unlock $100 million in unclaimed benefits in parents' pockets. If you're interested in that kind of ROI or helping parents stay out of poverty, come find me and my team in the back corner over there and we hope you'll join us in closing the family benefits gap.

Thank you. Let's go on a road trip. All right. It's no surprise that one-size-fits-all traditional career assessments have been outdated for a long time. When my friends and I were graduating from school, we went through a bunch of the old-school traditional assessments, but they felt that there was too restrictive approach to our future. So instead, we bought an old beat-up motorhome, painted it green because the green paint was on sale at Home Depot, and took a road trip across America

[01:44:00]

Interviewed people on a wide variety of a career paths to learn how they got to where they are today. People like a lobsterman on the coast of Maine, the designers and engineers behind Burton Snowboards, and the founder of Starbucks, Howard Schultz, all share with us their career stories. And by listening to people's career stories, it gave us a sense of hope and optimism for a future that we would have never gotten from going through a traditional career assessment. On that road trip, Forbes did a small article on our journey, which led to a book deal with Random House.

The book deal came out about a year later. It was heavily publicized by the Random House PR department. We were on the Today Show, Carson Daly's show, became the 15th best-selling book in America. And around that time, we realized that we were not the only young people in America who were freaking out about our futures, had a lot of anxiety, and felt that the traditional ways of finding your career were too restrictive. Because of that, we decided to pay it forward and build Roadtrip Nation as a way to empower students and learners and vulnerable career seekers all across the country to build their own road trip experiences.

For the past 25 years, Roadtrip Nation has been helping to launch these road trips. And today, we have a series on public television, 80 million households every year, our content's in schools reaching 18.5 million students, and we're in half of all U.S. prisons helping vulnerable, justice-impacted populations transition to meaningful career pathways. What most people don't know is that along the way, Roadtrip Nation's also built one of the world's largest media databases of 13,500 career videos of people talking about their workforce pathways.

We've always dreamed about mining the incredible ocean of metadata in these journeys, and it's not just the tactical stuff like what are your jobs and where do you want to go to college, it's vulnerable metadata as well in terms of who was first in the family to go to college, who was a first-generation immigrant, who had ADHD or a learning difference. But the barrier to entry was too high from a technology perspective. When GPT-3 came out, the barrier to entry went way lower, and we realized that since we had two decades worth of closed caption files on all of Roadtrip Nation's media assets, we could feed those into the AI and create what we'd always dreamed about creating.

[01:46:00]

Personalized career navigation experiences for anyone that didn't have the luxury of living in a motorhome and going across the country and doing it in a huge, giant, green motorhome. So that's essentially what we've been doing for the last 18 months. In partnership with the American School Counselor Association, which is the largest school counselor association in America, we've worked to co-design a virtual road trip prototype and implement the pilots across the country. We've learned two critical things along that learning journey. First, American school counselors have no extra time in the school day.

They have 700 to 300 to 1 counselor ratios, and if you're going to add value to their day, you have to replace something. And the second learning was that the one thing in their experience that was ready to be refreshed was the traditional career assessment. Students across the country are pretty freaked out right now, if you haven't noticed. Things are changing very quickly, and counselors don't feel that they're set up to help students retain that sense of hope and optimism for their future.

The virtual road trip is a five-stop multimedia AI-powered career assessment that guides students in exploring pathways for their future based on two decades' worth of Roadtrip Nation video assets, and guides them in building a real plan with counselors. We feel that this could be an Oregon Trail-like experience for career navigation at a moment of rapid change for students. For example, a student in one of our recent pilots who's gone through the virtual road trip, she hoped to be an engineer.

Her family was from South America. She'd never met an engineer before, let alone an engineer from South America. On her virtual road trip, with the vector search, the virtual road trip connected her with Catalina Laverde, a back-end software engineer at Spotify who grew up in Bogota, Colombia. Her family had to sell their childhood home in order to pay for her college degree, and after that experience, it not only went on with the story, but it also helped her with RiasTech science-based career assessments, career planning, and the students said, I didn't believe there was a place for me in the future of work, but now after I saw a story of someone like me, I believe there's a place for me.

We've been doing a number of

[01:48:00]

research-based assessments with UC Irvine and Julia Freeland Fisher from the Clayton Christensen Institute that show that AI-powered virtual road trips are helping to shape students' sense of occupational identity for their future. ASCA, the American School Counselor Association, asked us to be their keynote presenter last month in New Orleans. Previous keynote presenters were Michelle Obama, John Legend, and we were their day three keynote presenter in front of an entire New Orleans Convention Center of counselors who afterwards mobbed our motor home, and we're clearly so hungry for this experience.

And today, we're excited to announce a groundbreaking partnership for the virtual road trip. The number one career and college readiness platform in America, Naviance, that reached 40% of all public high schools, is this coming academic year replacing their traditional career assessment with the virtual road trip. In closing, we're trying to redefine what career assessments mean in America. We're looking for $2.5 to $5 million to roll out the next phase of the Roadtrip Nation virtual road trip to particular states and look at new categories like workforce pathways and post-secondary pathways.

We'd love to connect with you to think about what the future of assessments are for students across the country. Thanks. We have two more seconds to go. Accion, Accion, come on up. Hi, I'm Myles, and we are a bit of an outlier here today because I won't talk about upskilling or job placement. Today, we are solving health supply chain challenges in Africa to improve income. And by doing so, improve patients' health. About a year ago, I sat behind the counter of a pharmacy in Kenya.

That shop isn't a pharmacy the way you would picture it. For most people, it's the first place they go to when they're sick, and often the only one they will see. Seeing a doctor in Kenya usually means going to a hospital. So the person behind that counter is the frontline of care for that community.

[01:50:00]

And most of these shops don't even have a pharmacist in them. Kenya has 10,000 pharmacies and only 1,300 pharmacists. Additionally, 4,500 pharmacy technicians, which leaves more than 4,000 shops run by someone with neither qualification, making decisions about what to give to patients when something is not available. And this is how they run their shop. A stock card filled by hand, one line per delivery. Nothing is digitized, nothing is connected. And every few weeks, all of... And every few weeks, all of that paper has to turn into an order.

A PDF with prices on one side, a desk calculator on the other. I'm not shitting, actually, like these old school thingies. The ledger's in between. What will I need? What can I afford? And which of these actually earn me anything? Roughly 3,000 products on the shelves behind him and 9,000 on the national registry. When something on his shelf isn't available, the replacement has to come from that list. He's constantly interrupted by patients. I counted, seven times in the hour I was there.

He's also calling other pharmacies in the area in between. Some of his stock is about to expire in a couple of weeks. He's offering it for a discount to anyone willing to take it, then buys it back from wholesalers at full price. All in, two hours that afternoon, he told me later. This isn't a job anyone can do by hand and succeed. That same blind ordering that leaves that one shelf empty and about to expire, leaves another one empty. 88% of common...

[01:52:00]

Kenyan pharmacies run out of stock at least once a month. And when a medicine goes out, our own data across facilities show that it stays out for about 75 days. So on any given day, something a patient walked in for isn't there. Not because it's not in the country. Because for a pharmacy owner ordering by hand, there's no way to see what products matter most, how to utilize the budget best, or what could stand in if something is unavailable. Simply put, there's no way to plan ahead.

And for a shop with two employees, that isn't a line on a profit and loss. That's their livelihood. So here's what we built. You send us whatever you have, however you have it. Short forms like Zulu, acronyms like ATM or ABZ, the drug name spelled wrong, an Excel sheet or an expert from a point of sale system, even a picture from that lecture that I showed you. We love receiving structured data. The reality is it's chaos. Our AI reads that chaos, maps it to the national registry.

So when something's not available, it knows what can stand in for, then works out what you will need and what your cash can carry. That's running now across Kenya and Nigeria. What it can't see is price and sudden demand shifts. We want pharmacies to upload the price list they already have from wholesalers in addition to the inventory and their budget. Put that together with stock and demand data across the region, and the tool can tell one shop what to order, how much of it, and from what vendor.

Even factoring in demand searches from pharmacies across the region. Fully automated. We are asking for $550,000 for a patient that's their medicine being there on the day they need it. For those 10,000 shops in Kenya

[01:54:00]

And the far larger number in Nigeria, it's an estimated 20% more income. Not from working harder, not from hiring more, from not losing the sale. Help us give time and care back to patients who depend on it and generate income for people who need it along the way. If you're interested in some insights, visit me up there. Thank you very much. All right, I want you to give one more round of applause for our last presenter of the day, certainly not the least, the Foundation for California Community Colleges.

Hello. No pressure being the last one. It is such an honor to be here with all of you, to have listened to all these presentations. This is an incredible room to be in, and I feel like it is such a privilege to be in front of all of you. So thank you. My name is Kavitta Ghai, and I am the co-founder and CEO of Nectir. And I brought my very technical CTO with me to help me with the slides. When I first got to college, I spent my entire freshman year wanting to drop out.

I am autistic, I have ADHD, and I'm a first generation college student. So no classroom that I've ever been in before in my life has felt comfortable for me or my brain to be in. And then I got to college and I started paying $40,000 to be really uncomfortable. And that's finally when it became enough of a pain point for me that I said, either I'm going to drop out or I'm going to do something about it. And so we founded Nectir, the safe AI solution for schools.

And we've now partnered with the California Community College system to bring safe, guard-railed AI to over 2.2 million students across the state, all of whom are asking some version of the same question that I had. What do I do with my life? How am I supposed to do that?

[01:56:00]

And most importantly, what's my next step? Now, for most students, the problem starts way before you get to college. For me, it started in high school when I got into a bunch of schools and I ended up picking UC Santa Barbara because it sounded fun to be near the beach. That was the entire reason. And it was. It was super fun. I loved it. But now, in hindsight, I know that what I should have done was think about what career path I wanted in my life, what jobs aligned to that, what major would get me to that job that I want, and then pick the school that matched all of that criteria and fit my budget.

But I had no one telling me that at that point. And I had nowhere to go research all of that. I had nowhere to see what my options were. And that's exactly what we're solving for today. For me, the biggest issue in school was the fact that I had no idea what I was there to do, what I was there to study. I knew I liked business and I picked a school without even a business major, let alone a business school.

So I ended up picking econ and accounting first, very quickly realized that is not the business that I was thinking of. And then I picked technology management. That wasn't it either. And then I picked communication because it seemed like the easiest thing to do. And then by my senior year, I dropped out with one class left before I would finish my degree because even four years in, I couldn't see the light at the end of the tunnel. I had no idea what I was going to go do with a communications degree, and I had no way of knowing if I could transfer all of those units that I had already gotten to another business school.

So I gave up. And it might sound crazy, but that is far closer to the norm of what most college students in this country experience. Of the 2.2 million students in California

[01:58:00]

California's community college system today, less than half are going to graduate with an associate's degree or transfer in six years. That's supposed to take them two. That in itself is the entire problem, and we now have the power to solve it. For the first time ever, with the power of AI, we actually can show students every single one of their options before they choose their path forward. We can show them all of the schools that are available to them, all of the different job paths, and that's exactly what we've built with MapperGPT.

For the first time in California's history, a student can now get real-time support anytime they need it, even before they've enrolled in the system, to figure out their degree pathways and transfer requirements. And there's already 102 participating colleges across the state. So now any student, or anyone that even wants to go to school in California, can go to the California Community College's Program Pathways Mapper website and have a conversation and get a personalized plan within minutes. So let's take a look at what it would look like for me if I had this resource.

So first I'm going to go to MapperGPT and let it know that I like coding, and I know I want to stay near home, and that's literally all that I know so far. And what it's going to do is query across millions of disparate data points across the entire state of California that we've unified into one coherent system that speaks the language that students understand. So it gives me the top three options based on exactly what I'm looking for. But now I want to take it a step further, and I want to know which one pays more, because now I know that's what I should have been asking.

And so it's going to also pull localized real-time employment data to show me what the different salaries would be for these positions that I might get from those majors, and it can even tell me which one of these jobs won't be replaced by AI

[02:00:00]

by the time I graduate in four years. And now I want to take it a step further. I want to know before I commit to taking this class at this community college, will those credits transfer over to another school in California if I choose to go there? And it does. We were able to map across those 102 schools all of the course equivalencies. So we know that COMB 11 at one school is going to match COM 122 one to one. You're not spending extra dollars and taking six years to finish your degree.

You're taking two. And the last step is to build a specific, detailed, personalized map of every single class that you should take over the next two to four years to be able to get onto that career path in time and know that you're not going to have to go further into student debt. And I can go back and change this personalized plan anytime I want as my circumstances change. This is what the future of going to college looks like. With over 20 million pathway views to date, California has proven that MapperGPT works.

And now it's time for us to scale this to the other 10 million community college students across the nation. With your help, we can get this nation more educated and more employed than ever before. And I promise you that all of our lives will be better off for it. If you believe in the power of education and AI as much as I do, please come find us and let's talk about how we make this happen. Thank you. Okay, so I'm going to wrap up really briefly and then we're going to go to reception, but I want to leave you with a couple of quick thoughts.

What a kind of walkthrough of the opportunity that is before us, what the future may look like just around the corner. We do these events each year. We might do them more often, but think about how much has changed in this last year. The next year when we come back together, the world may look like

[02:02:00]

is a very different place with the progress of AI, the progress of change in so many different systems. So I want to share a little bit of a different reflection that maybe is used at the end of these sessions. We talked a lot about the positive side of this work, but we've been along with these teams for years now, and we can tell you real stories of governance fights that have shut down organizations, of people negotiating contracts that have fallen apart and really made bad blood and had teams really tear apart and not be able to finish challenges, really challenging legacy integrations within organizations that are going through change management systems, that are learning how to do things in entirely new ways.

There are so many ways that these projects can be challenged to progress to the vision that they see before them. But together, right, we can navigate those with them. We can give them flexible capital, we can support them, we can help them with guidance and mentorship, we can connect them to other teams who have gone through those challenges before. But it takes both sides of those that are in the room today, right. These are teams that are often going through this the first time.

And really for a lot of us in the back of the room, particularly the funders, you're seeing a lot of these different projects. And so we're going to ask you as you leave today, connect with one of these teams, follow up with them, ask them how you can help and help. See if you can really do it, right. Connect them, find a way to support them. But there are only a couple things that we can be certain about, right. This world is changing so fast and I want to say just three.

AI models will continue to surprise us with their capability, even announcements today, right. These are going to continue to get better. It's going to be a vertical increase in capability. Second, projects like those today can improve real people's lives, right. In so many ways, we've seen it just the inklings, just that little bit of the silver lining. But that's not a default outcome, right. It's not a default outcome. That's also something we can be confident about.

[02:04:00]

that humans are super, super, super good at using technology for the worst possible reasons, but it takes intentionality and thoughtfulness and partnership and funding and patience and leaders like those that have stood up on the stage and been brave doing this for the first time to make that a true reality, a real future. So we want you to join them in that. So another round of applause for the teams, and we're going to wrap up for the day and let you go.