Kumar Garg: How philanthropy can fuel scientific innovation

Everybody, I’m Alex Kotran. I’m the co-founder and CEO of aiEDU. And we’re here for what is sort of like one of the very early recordings of AIE Edu Studios.

Our producer CJ is here. We haven’t even decided if we’re calling this a podcast or a show. By the time we publish this, we’ll probably have decided that.

But I’m very excited to have Kumar Gar a longtime confidant adviser funer and supporter of aiEDU and someone who has now very recently gone off and sort of shifted vantage points and become a founder in his own right of an organization called Renaissance Philanthropy. So we’re going to hear more about what Renaissance is all about. But before we dive into any of that and even giving you a chance to share your background, I was curious if you could why don’t you start by sharing a little about who you are very high level your journey.

We we you were at the White House, you were at Futures. But then I’d like you to land on, you know, what are you obsessed with right now? Great.

Well, great to be here. It’s great to see the great to see this work and great to see all the progress the organization’s been making. It’s really been fun to watch.

I think just to sort of quickly cover the highlights. So yeah, I was born in India. I was born in Delhi.

I came to the US in middle school. Grew up in the suburbs of New York. Studied political science and computer science back in college and got I caught the political bug.

So I I got the political bug right as I was getting naturalized. I was in on college campus when the presidential primaries were happening. So that got me the political bug and went to work for then Howard Dean when he was the anti-war candidate.

Seems like a long time ago. And had a you know talking about managing a team. I actually managed a team of 20 people as my first job.

So like I was the deputy communications director for New Hampshire for Dean. You know, it’s one of those things where when you first start like you’re doing everything and then the campaign grows around you and so then you know you’re like handling a whole team that is like sending out trackers on all the candidates that’s handling a candidate that’s doing all these press interviews. So it was really interesting to have that as your first experience.

The way I got into education was I was in graduate school. I was in law school actually and you know I I like law school but it wasn’t it just felt a little distant from like well how does this actually play out in the world and there was a clinic that was being started at the law school specifically to sue the state of Connecticut for underfunding the school system and the reason was the plaintiffs which were a number of low-income school districts in Connecticut were basically getting like left out of the school funding formula just based on the way it was getting it changed and they couldn’t afford like they had tried to do it a version of that lawsuit with private lawyers and run out of money. They came to the law school and said would the law school take the case?

Now it’s a kind of dubious proposition for law students to be doing complex litigation. So we found a bunch of senior attorneys from various law firms that supervised the case but it was a studentled case and I thought that was quite interesting. So I that was like my big area of work throughout throughout school where I would interview all these parents.

I would interview all these superintendents. I would try to build the actual case to say why is it that these school districts are struggling? How does finance and school funding relate?

And I spend a lot of time actually thinking about how you actually structure litigation to create political change because the you don’t actually, you know, win a lawsuit and then suddenly the school system changes. The school system is a way for you to shine a light on those deficiencies. And so it so we actually went up to the Connecticut Supreme Court on an emergency appeal because the then governor took the position that there was no right to an education for Connecticut students.

So we took that to the Connecticut Supreme Court and we won. So it was this is a Democratic governor. It was a Republican governor then.

This is you know politics change and there I mean they took a sort of a unconventional position where they said even though you might have a right to an equal education you don’t have a right to an education at all. And so we won in front of the Connecticut Supreme Court. It was right around then that President Obama was getting elected.

I was I just finishing up the case. The case was going back into district court. I had real interest in working in policy and I got myself a fellowship to come to the come to DC.

It was one of those things which when people now ask for life advice, I always tell them to think about policy fellowships as a way in. There’s nothing more powerful when you show up in DC to say you just need to give me a desk and a business card and a phone. Somebody else is paying for this fellowship to start.

So I was able to get a job in the White House Office of Science and Tech Policy partly because you know I’d done some computer science back in college. I’d just done this education case so I knew something about education and they were like okay you know maybe that’s enough and then basically they were hiring me as a generalist to basically say you have some political background some background in education you kind of learn the rest. So they were taking a bet on me.

I I didn’t realize you’d been So you’d been at OTP starting as a fellow and you you were there the whole eight years. The full eight years. And so when I first showed up I literally had no idea what policy was.

I had never served in government. It was a totally new experience. And I I would say the first two months I literally you know I had this sort of like sense of like did I make a big mistake?

Like I I don’t know what I’m doing and I feel like I’m failing. You know, I go to meetings, but I I have no sense of what it means to do this job. And the there’s a it’s a very steep learning curve.

The thing that I had the benefit of was I was working for the deputy director of the office Tom Coyle and he had this apprenticeship model where he would basically say none of this will make sense but the easiest way for it to make sense is just come with me to meetings and before the meeting we’ll talk about what the meeting is about and after the meeting we’ll talk about what happened in the meeting. And what was fascinating was all meetings actually play out at two levels. There’s like the formal reason why you’re meeting, but then there’s also the subtext.

Like what is it that the different people in the meeting are trying to accomplish? What are all the side conversations? What is the the way policy actually gets done through the workings of like the agenda?

Who’s there? What is the larger policy process? So from that moment of 2 months in, I’m like this was a huge mistake.

At the six-month mark, I had done two events with the president. We had launched a major initiative on math and science education that the president announced. So it was the first major science announcement that he made and it was all about how do we get kids more excited about math and science education.

And then that sort of actually like set the stage for my why I ended up being at OCP for the full eight years where you know I was the first it was one of the major early wins for the science office to have this major initiative announced by the president on math and science education and I had been the person who had put it together. And so then every time we wanted to do the next science and tech initiative and pitch it to the West Wing, you know, they would say, "Well, you know, you did the last one. You know what?

What?" and so I sort of became the the the person who got added to every policy initiative that could get presidential level attention. So even though I had been originally hired with this very narrow specific education job, you know, education became sort of my one-third job and two/3s were everything else that was moving through the office. So I got a chance to work on advanced manufacturing and space commercialization and the future of the internet and then the policy actions we took after healthcare.gov and everything else.

So it was you know we ran this tech and innovation team with this really wide mandate. And so by the time I left, you know, most people spend two years in the White House. I spent eight.

So it felt like this like, you know, from newbie to greybeard where by the time I was leaving I was like, "Oh, like it’s very hard for you to, you know, you could pick any topic." And I’d be like, "Oh, you should here are the five people who work on that. Here’s how you would construct it. Here’s how, you know, we would take this from an idea into making into initiative.

Here’s how you’d get the agency involved. Here’s how you get philanthropy involved." and it you would not find that description of that’s what policy is in any book, right? You sort of people sort of imagine view of policy is that it’s a knowledge production exercise.

You think really hard, you come up with the right policy answer. You there’s some mechanism by which it becomes policy. That’s what policy makers do.

But they the lived exercise of what policym is is actually to actually think really hard about what’s not happening. Come up with a list of those things and then figure out what’s stopping that from happening. Like actually this agency needs you know additional support and doesn’t want to boss this other agency around.

Your job is to take this agency’s good idea and give it to that other agency because you know there’s dynamics there’s equity dynamics between two agencies or in this instance this is actually an area where there’s really you know when we were building the president’s brain initiative the key insight behind that initiative was that the technology around being able to you know understand the brain was technically improving a lot where for a long time you either did full brain pain scans or you studied individual neurons. But the idea of studying hundreds or thousands of neurons and how they interact with each other was just becoming possible. The idea behind the initiative was this could become a powerful new platform for how we actually do brain science.

You know, that actually opened up a whole bunch of new ideas for partnerships with medical technology, with the cloud companies around how you might actually store large amounts of data because of the way this would work, with a whole bunch of different patient groups around diseases you could go after because of this. And so to actually build an initiative, you actually might do a bunch of things that are not like I’m just changing the rag from here to here or I’m just adding this as a new priority in the next NIH call. So that was interesting.

It was like a you sort of learned that in any given moment, you know, America’s a big place. There’s a living breathing set of body of work. So if you’re going to figure out what the right policy answer is, you have to be able to actually first figure out what’s like what’s happening right now, what’s not happening that people think would be a big deal and then what are the what’s the most useful thing government could do?

Is it that it can convene or is it that it can fund? Is it that it’s actually the blocker? You have to actually change the rag.

Is it that it can actually make something feel more possible just by stating the goal? So sometimes it’s just saying this is now more possible and shining a light on it and then everybody sort of says oh you know this is a much more is it that it you know provides itself as a customer. So one of the big things that allows DARPA to be so successful is it able it’s able to find internal government customers for its innovation.

So I did that and then you know when I left I started to explore well what does it mean you know if you’re working across all these science and tech topics what does it like mean to not do that for the government which was kind of a journey but one of the things I ended up doing was then working with Eric Schmidt who had just finished as being executive chairman of Google was interested in science and tech philanthropy and was starting to build a team. So I work with him for just over six years. And then Renaissance Philanthropy is just the most recent, so less than a year old.

And the real idea is that Congratulations. Yeah. I mean, it’s been it’s been really interesting.

The idea behind it is how do we how do we take this moment where there’s actually an incredible amount of excitement about science and technology but and actually build an infrastructure so that philanthropists who are interested in these topics can actually bet into those topics. People find the topics exciting but very hard to actually do larger scale bets in them. And one of the reasons is that the there’s actually not good mechanisms for it.

So if you might be interested in a particular topic and you might say well why aren’t we making really fast progress in cancer or why aren’t we seeing more alpha fold like breakthroughs on AI and you know most of the time you might say okay well I I’ll talk to a couple of academics on this topic and they might be actually really smart. So they’ll say you know you should fund my lab and so you know the person might say okay I’ll fund your lab but then two or three conversations in you know most folks who can support this kind of work you know they have busy day jobs and so they’ll say okay I’m sort of funding three people I’m like going to go back to my day job but actually the amount of capital they’re deploying compared to what their potential to give is minuscule now you might say to yourself well why don’t you deploy more, but you know, these are still folks who want to deploy it into a strategy. Who’s going to put that strategy together?

So, you could actually fund a range of topics with a goal in mind. So, one of the ideas that we’re trying to popularize is this idea of a philanthropic fund. So instead of donors betting on individual projects, just like you do an investment fund, why not you build a timebound philanthropic fund that says, "Okay, the goal of this fund over the next 5 years is to move this space from here to here.

I’m going to make I’m going to I’m going to be the LP the way you are in the investment world, but I’m going to do it philanthropically. I’m going to put money to work into that fund and then that fund is going to work across dozens if not hundreds of researchers to make a series of whether it’s research awards, whether it’s catalytic market activity, whether it’s policy work with the fund goal in mind. And so that that was a sort of sort of like question which is could we build these funds as a way to give donors a ways to deploy much more many more resources but with still strategy in mind.

And so it’s been it’s been exciting. So, you know, we’ve been, you know, less than a year old and we’ve already launched three funds. And so, you know, people when we were just coming out said, you know, I got a lot of feedback, but the biggest sort of feedback I got was just that it, you know, like the advice was almost like good luck.

You know, it’s like fundraising is really hard. Donors are obsessed with control. So you’re not going to, you know, why would they put money into a fund where they don’t get to make all the decisions?

And you know, the thing I say to people is that you’re just seeing the part of the iceberg that’s above the water. So your experience of a philanthropy are active philanthropists with active program officers who are actively managing just like you would meet a DARPA program officer. They’re opinionated, but mo but most of the the actual philanthropic capital is below the surface.

It’s dry powder. It’s people who are busy doing other things who don’t have the time to actually build out a strategy. For those folks, you actually need to give them much more pre-built vehicles which they can deploy into.

And they actually don’t want the control. They actually want to bet on a strategy and they actually want to trust the you know the person who’s going to be the fund director to drive that strategy. So that was the the sort of broad thesis as to whether you know those kind of vehicles would be important and it’s a little meta but like you know if you sort of think about it’s partly why we named Renaissance Renaissance which was we sort of hearken back to the you know most of the science and tech organizations always hearken to the 20th century so it’s always like Bell Labs or you know you know the space race but we said let’s think about like you the scientific revolution that started, you know, with the original Renaissance.

And one of the things when you read back to that period is that patrons actually played a really powerful role, which is just backing individual artists, individual entrepreneurs, and individual scientists in being able to do exploratory work provided a lot of the risk capital to get that area going. And it provided a flourishing even as we you know that was necessary in that period. And so this idea that the patron can play this really powerful role.

They don’t have to direct the work. They can just, you know, they can say, you know, I think this body of work is exciting. You know, you’re in charge.

Why don’t you figure out what’s next? That sort of patron model, I think, has fallen a little bit. But it’s in, you know, it’s part of our story and, why not sort of hearken to that?

And you know, and then the other analogy I always give is to LPs, which is that a lot of folks totally understand in the investment side that if you’re trying to deploy large amounts of capital, you don’t do thousands of direct deals. You know, it’s very hard to do that at scale. You deploy into funds and then they make the decisions.

And so from either perspective, either from a finance kind of deployment perspective or from the notion of like what does it mean to make bets, you know, having additional ways of doing this work in science tech becomes exciting. So that’s been a little bit of the way we’ve been doing it. And so, you know, for example, we launched this fund specifically on AI and math.

So, you know, AI is in the news a lot. Part of the reason it’s in the news is that every week there’s a new announcement and the new announcement says there’s this new model this new model is showing additional capability on general reasoning. Isn’t that amazing?

And what you know the term sheet that comes out attached to every model is like here’s all the public benchmarks and here’s how it did on those benchmarks. The implication being that these models are going to be automatically used now in the real world to do stuff. You know, one of the questions you might ask yourself is like, well, who’s doing that work?

The who’s implementing it? Who’s actually taking the the capability and then running all the experiments to then use that capability on something that really matters, right? Oh, we’re going to take this these models, we’re going to mix them with satellites and we’re going to try to see if we can detect file wildfire within minutes of it starting or we’re going to take these models and we’re actually going to make field metalists like Terrence Tao more productive.

So that’s what the AI for math fun does where the idea behind it is not like at a esoteric level but like what is the workflow of actual working mathematicians? Are there ways you could actually build extensions of these tools that could actually make allow hundreds of mathematicians to work together on a problem? You know, and so, you know, there’s actually a lot of ideas.

Those are not necessarily commercial ideas, right? Because it’s not like the world has millions of mathematicians, but from a public and social goods. If you could make working mathematicians more productive and they could solve the underlying theorems that then power the internet, power cyber, would be a big deal.

Yeah, it feels like a good fit for philanthropy where there aren’t necessarily like obvious market forces. I mean this is this is a brilliant model. The the one of the things, you know, as a as a as a founder of a nonprofit, so sort of on the other side, and whenever I talk to my peers, I think it’s always it always surprises me how much dissonance there is between, you know, the nonprofits that that they feel like there is sort of the scarcity of capital, right?

And when you talk to funders, they almost feel the same way. I mean, many times funders feel like we’re really struggling to find places to put our money. And I’ve talked to some some of the biggest funders in the country where they they did not even meet their targets.

And and I think that it always surprises, you know, some of the executive directors that I’ll that I’ll talk to. And it’s it’s just it’s it’s easier said than done. Because nobody actually wants to just sort of throw money out.

And accountability is hard, right? If you don’t have the expertise to really be able to you know to validate impact because obviously you know in the social sector there’s a lot of massaging of impact and you know the need for someone who can really sort of like see through and ask the right questions but the other thing that strikes me is you look at some of the biggest foundations and I won’t name names but if you look at some of the biggest foundations in the country the institutional foundations you know their their fortunes came from this it’s very old 20th century sometimes 19th century industries and so it is interest it’s it’s sort of fascinating this idea of this new crop of you know millionaires billionaires who you know they wrote out of you know ’9s 2000s really when venture capital like like really birthed as a meaningful if not like huge and massive source of of of investment capital. So it’s sort of a model that makes a lot of sense to the sort of newer category of you know high net worth individual that you know where sort of older institutions just don’t have that sort of like those that you maybe the interest or sort of just the experience with it to really make that sort of like the connection to the model that you’re describing.

Yeah. I mean I think one way to think about it is you know we’re always heroes in our own story. So it’s it’s for the for the the researcher with a lab, for the nonprofit leader, for the social entrepreneur, you know, they’re like, "We’re doing all this great work.

How do we go out there?" The same thing happens on the donor side. I think the the distinction I would say is that I think you have to take philanthropy more seriously as a sector. And so one of the things I think if you think about as a sector is that people don’t actually know how philanthropy ended up operating the way it operates today.

If you sort of go back, so open philanthropy did these nice deep dives where they said what are the 10 to 100 best successes of philanthropy, you know, is like someone actually tracking what are the out if you were thinking about this as a venture capitalist like you know what are our thousandx wins for the sector and can we then use a reference class analysis to say well what are the underlying components of those wins that then when we’re building programs we should build and the funny thing because they did a they tried to do a lit review and they were like this like not been done. Can you name is there any examples that come to mind cuz I’d be very interested in that. Yeah.

So they so basically so now they put this up. It’s actually a really good resource. I mean they did they basically read so there’s individual accounts of famous philanthropists like oh here’s how like the early days of the Ford Foundation or Rockefeller’s work with the green revolution or you know the work that Yeah.

You know like so you know malaria I’m guessing yeah Warren Weaver did with you know early days of molecular biology but it’s they’re pinpoint but no one had actually tried to put it together so they they took one attempt to basically trying to say what are the biggest wins and what do we actually know like you know one thing they found is some of the biggest wins have a policy component so that that what one of the things that those wins have is that they were enabling a policy change that then sort of ricocheted and got scaled by government. That was sort of interesting. Second is they were often taking neglected topics, you know, things that parts of the research community believed but were not mainstream and and sort of really helped them.

So there’s there’s a number of different components that they sort of found as like ways to sort of approach it. But what was interesting for me was that one that that research literature did not exist and two it’s not that if you went out into the sector and you were like you know surely you know what are the biggest wins of the sector that this was just sort of base knowledge. So where whereas in venture capital you were to ask like what are you when you’re making deals like what are you trying to emulate?

They would like immediately whatever their you know vertical is they’ll be able to tell you like the 10 unicorns that they are sort of like looking for analoges of looking for the next Uber or the next you know Slack or like you know platform play in this specific SAS area right so what are your comps right and so one of the questions I always say is like who are the 10 best program officers in America right like is there even a list like that there’s not right but like but program officers that work at foundations are capital deployment experts You know, as someone who’s like worked now in this area, I have my I have like my list of favorites of people who I think are incredibly good at spotting things before they anyone else. They’re good at using money to get those ideas going. They’re just like really good at it.

But I’m like, you know, I have like a sensemaking way of doing it. But I’m like, these people should be heroes. Like they dollar for dollar are creating, you know, often 100x impact for every dollar they’re putting out on the street.

You know, you look up look them up on the website, they just sound like anybody else, right? So whereas like, oh, I was the early check in Uber, you know, that’s like people will put that in their in their bio. So so one is I think there’s not as much understanding of like how did the sector end up, but the second part is that how the sector ended up where it is is historically contingent.

So I think that and so my view is that because the early wins all came from the professionalization of a dedicated staff. So if you sort of go back and you say where does philanthropy come from? It comes from things like the Ford Foundation, right?

And then the Rockefeller Foundation which were the pretext to then the Gates Foundation. And all of those were efforts to say if you build they were basically efforts to and then the government was doing the same thing with the creation of DARPA and NASA and others. It was the idea that if you build core professional teams of experts, those experts can go out and build strategic depth and then they can be smart deployers of capital, right?

And so every person who’s come since when they say quoteunquote get serious about philanthropy what they sort of go down the expected path to go down is you build a strong professional staff and that staff makes decisions. This is where the arrival of the program officer comes from. Now if you just flip the coin and you say well how did investing go?

Right? So this also capital deployment but how do how do we invest money you could actually say well actually investment capital in the US went down a different path right so in the ’ 50s and 70s you start to have these large pools of capital that need to be deployed it used to be they were just sitting banks but then you start to have the emergence of these expert deployment engines the the rise of the venture capital industry the rise of the private equity industry the rise of hedge funds, the rise of all of these intermediaries that say we’re going to become sector experts. We’re gonna build brand that we’re really good at what we do and then we will go to the endowments, the pension funds, the the family offices and we will say we’ll put your money to work and here’s going to be the and so now that system that now you know you’ve played that out 30 40 years it’s like to direct investing is considered like a risky you know like bold move.

If you if you’re a if you’re a limited partner, you might say, well, we mostly invest into funds, but you know, we do do some direct deals because we want to, you know, see the work up close and all that stuff, but you have to kind of justify it because people would say like, why are you doing direct investing? Of course, you should invest through funds. And I in my mind, these are just two sides of the same coin.

Why is it that in philanthropy direct investing is through your staff is considered a smart move. That’s what that’s what it means to do serious work. You know, everything that you do through intermediaries, collaborative philanthropy, pulled funds, everything has a bad name to it.

It’s like considered high transaction costs, not particularly motivated. You know, you know, most people don’t do it. They they want to do it directly.

And then on the investment side, it’s the exact opposite. And my theory is just you you basically have these two sectors running in parallel developing different notions of what it means to be serious. But both sectors can borrow ideas from each other.

You know, in the investment world, you start to have now people saying, well, you actually need to go preede. You actually need to build studios to actually find ideas that are not yet commercially viable. Like you need to get more active.

You can’t just, you know, you can also have philanthropy start to say, well, you can actually have expert deployment. You can actually have funds that are led by actual field experts that are time bound that you can park money in. So, in my mind, this is like there’s no one right answer.

It’s just you need to actually think of these as basically in you know on a risk spectrum. And so you have to like think of it as a sector and say what is it that this sector is missing that it can learn from other sectors that are also in the business of deploying money and say well why don’t we have that like is it like is it that like the world of professional investing is totally off like like why is it that everyone uses those and I think you know it’s not that they’re off it’s they’ve actually figured out that there’s huge value to specialization right like you can actually build a fund that’s very specialized in a particular part of the world or a particular sector or a particular thesis, you can be world class at that. And you know one of the challenges always is in philanthropy if you want to have many different interests you have to build ever larger staff and that’s hard.

Not everybody wants to have that. So, can you develop models that are flexible in allowing you to have like, you know, what I consider more pointy bets, but to have many of them, right? And currently, people are like, well, you should focus.

It’s like, you know, you can actually it’s it’s not a question of you focusing, it’s a question of, do you have ways to make focus bets without driving yourself insane? Right. Right.

Well, I think you’ve answered the question of what you’re obsessed with implicitly. Yeah. No, let’s let’s let’s keep going down this this road.

I I have a few I have a few reactions. One is I think the nonprofit sector is similarly struggle struggles with this the dissonance between sort of like how startups operate and there’s like sort of good and bad aspects to it but there’s a lot of like good aspects to the startup model much of which is like relatively new and recent and you know the nonprofit sector just doesn’t have the the the muscle memory and in some cases just the interest they don’t they I think there’s almost a a a rejection of this idea that well we’re not a company like we’re trying to do good. And I think in in founding AID and I was lucky to be a part of you know a couple different nonprofit accelerators that were basically designed to say okay like let’s take let’s borrow the best aspects of startup strategy and try to meld those with sort of like mission and impact driven work right and you know I think one of the one of the things that I found is you know nonprofits are expected to you know skimp on hiring and pay lower wages and and and funders are many especially you know sort of certain funders right are they they they’re obsessed with like you know restricted funding that’s like very time bound and you know in the startup world it’s it’s the opposite.

It’s like no we need we need flexible capital. We need the ability to be to move quickly to sort of like identify opportunities and like take advantage of like sort of the arbitrage and being that first mover. I think there’s what you’re describing by the way I I I also recognize that for like deep science I assume a lot of that is actually restricted investment and things like that like but in building enterprises and building sort of like a you know like 51c3s there seems to be this growing recognition from funders that you know unrestricted capital while it may not be the only way that that that we want to fund it it it’s really important for nonprofits to have access to that.

The other thing that I want to this is just a comment. Yeah I mean I think the the piece on that that I would just say is that I always you know whenever we do like meetings with grantees or with just you know with partners in the field like I’ll of let’s say it’s an unconference I will always do like one session on like revenue/f fundraising and I always get a ton of people that come because I actually think it’s a thing that every nonprofit leader worries about obsesses about how are we going to grow, how are we going to stay alive. But that there’s actually very little training or even mental models for how to think about it.

So one of so for example, one one mental model I give people is I say to people what do you know what your peers do? Like at a very basic level, I’ll say, "Okay, you aspire to be a nonprofit that is operating at a 25 million a year turn. You your revenue across fundraising, earned revenue, sponsorship, everything is 25 million." Like, have you looked up the 990s of organizations that are at that size?

They’re like, "No." And the reason I always tell them to do to look is that one of the things you realize is when nonprofit leaders are early, they will have the wrong lesson they will pick up from startup world is this idea of product market fit, right? So they’ll say, you know, like I just need to find something that is working and if it works, I’ll just scale that. You know, if you’re trying to sell a product, you’re like, "Okay, this product, people are willing to buy it for for a price higher than it cost me to make it.

I’m just going to make more of it." You know, it’s not a bad like basic structure. But in the nonprofit world, lots of things that constitute revenue have like a risk variable attached to them, which is if you just overindex on that type of revenue, you are actually putting the organization’s like kind of one, the overall amount of ability to operate at that level is going to be restricted. But two, it comes with a question mark.

So, you know, this happened like I did a lot of emergency giving during co, you know, we were like trying to get resources out on lots of important ideas and I have I had all these instances where they would be like somebody from Silicon Valley who’s like wants to do something important like, hey, we’re going to we’re going to help people find vaccines. We’re going to quickly stand up this organization. They’re used to raising like venture capital money, quick round, everything else.

And so, you know, I gave them resources quickly. We’re in the midst of a crisis. And then they would be like, "Okay, like I that seemed to work.

I’m just going to keep doing that." And I would be like, "No, no, no. This is rare. Like you actually need if you want this to be just a short-term sprint, that’s fine. But if you want this to be an organization, you’re going to have to build like you’re going to have to figure out what your peers do.

And what your peers do is they actually build a bunch of different things. They develop relationship with the government so they can get grants and contracts. They develop revenue streams where, you know, they’re able to like they run a center of excellence so they train others.

They’re able to like have a core of what they’re working on, but they’re able to then build revenue in a lot of different ways. And the experience I always have with nonprofit leaders is that they will treat whatever they have figured out as very straightforward. Oh, you know, we know how to raise money.

We just we just go meet with donors. We explain what we’re doing and it works. And then I’ll say, "Oh, have you tried to ever get a government grant?" And they’ll say, "No, no, no. You know, that’s like too hard.

Government so unpredictable." Obviously, that’s even more true with the way that the past few weeks have gone, right? But I’ll say to that same nonprofit leader, okay, like just like go on LinkedIn, find someone who used to work in that area that knows how to do government grants and like have them talk to you, you know, like hire them as a consultant. And I’ve had instances where then that same nonprofit leader two years later will be like, "Oh, we are now like a major provider of services and this government agency gives us six different contracts to do that.

And so they went from like that is totally impossible to like this is actually a pretty core part of how we keep the lights on. And the only difference was that just the willingness to say one we need to have we need to try across multiple things. We don’t we can’t treat the things we’re bad at as black magic and then you know and an effort.

So just as like a little tangent that the the piece what makes it different when people are doing it on the nonprofit side is that you know it’s almost like diversification has this huge extra value because if you can actually grow and so that’s why I always say look up the 990s. You can’t find a nonprofit, there’s a handful that once they hit a certain size, if you actually look at their cost structure, they’re not having multiple different forms of revenue. It’s just it’s the only way to do it.

So, you have to ask yourself, are we special or if we aspire to operate at that scale, how are we going to get there? Yeah. It’s I mean there’s a whole this whole conversation about grants and government grants and and but what you said about policy I think is even connected to that because I think especially in the education space where you’re there’s there’s very few federal levers.

I mean there’s funding but in most cases you’re going state by state district by district even within states like with local control like Michigan or Ohio. And it’s it is a very arduous path to getting because you can get one grand, but if you were like trying to consistently get you’re trying to get into the budget and try to like get part, you know, plugged in with some of these like, you know, local institutions that have 30 50 years sometimes of relationships in the state house. I think what’s intimidating about that is there is no shortcut.

There there usually isn’t just a consultant or a lobbyist you can hire that will open door. You know, there are certainly steps that you can take and we’ve done that in states like Ohio and Washington and and others. But I think I think part of our success has been this, you know, I don’t know that I thought it as like diversification of revenue models because we’re like right now highly biased towards just philanthropic capital.

But I think one of the things we’ve done is we’ve invested in sort of long-term relationship building and ground game. That has begun to unlock, you know, big state grants in Ohio, for example. It took us almost four years to get to a place where we’ve gotten like now two almost half million dollar grants back to back.

But I but I think the distinction I would make is that I think you’re curious about that and exploring. And I find that a lot of times folks who are building social enterprises, folks who are building nonprofits don’t you know you have to be curious around how to put it all together. So one example I have is like when I was working in the White House, you know, lots of folks would come, you know, be like, "Can I get coffee or can we can we tell you about our organization?" You know, that’s part of the job.

I would meet with lots of organizations working in education elsewhere. Most of the time, like if I met with them a year later, they’re like, "Oh, we’re coming through DC again. Would love to meet you." You know, they would say, "Oh, you know, we were working with 10 school districts.

Now we’re working with 12 school districts." And I remember I had this distinct memory where like every year I would meet with project lead the way and you know year one I would meet with them they’re like we’re in 500 schools and then next year I meet with them they’re like now we’re in,000 schools the next year I would meet with them and they’re like we’re in 1500 schools like like what is it that you are doing? I mean to be clear, I know in the venture capital world that would be seen as sort of middling performance, but in the nonprofit world as incredible. Yeah.

Yeah. So agree. But I guess what I’m saying compared to the reference class I had, which is like very modest improvement in cap in capturing the market, they were getting, you know, they were like consistently growing in these like, you know, year-over-year turns.

And I said, "Oh, like what is it that you’re doing?" Right? And so they were like then very you know they had a what I thought was a pretty interesting strategy. They said oh you know we actually don’t use philanthropy at all to run the core of the organization.

We just use philanthropy to buy down the price of the product. So they were like we price the product at what it costs us to deliver it at full cost and then we go to philanthropy and we say we would love to provide it to this school district. Some school districts are going to pay for the full, but in this district there’s a lot of need, there’s a lot of opportunity.

You can buy it down. And they said, "What that means is that, you know, the up and down of our various philanthropic asks never affects our core. It just affects our coverage." And they were able to keep growing it out.

And I was like, "Oh, that’s a never heard of that model." And then the the second thing they said was they said, "Oh, we we buy into the sales market." So they were like we went and we said well how do textbook publishers get their stuff out to everybody and they said oh it’s actually like there’s a set formula you know for how do you actually it’s a sales channel just like device manufacturing and others and you can tap into it and for me what was like what was interesting was I was like this is like really valuable information that are sector insights how do you scale right people always are like how do you scale ideas in education how do you scale ideas in any sector and Those I think often exist but like we don’t do a particularly good job of actually saying like here are the people to learn from who are figuring out the whereas like you know other sectors people are like oh this is you know if you were in the startup world you’d be like oh when you’re sort of raising your series A series B you get a sales team the VC funds you know the VC firm that’s investing in you says what given where you are here’s exactly the your next growth path involves this way of growing out you know a sales team to then get this much market capture and I think you could bring some of that you know obviously it’s not a perfect analogy but you can say like what are the strategies that people are using that actually are working for them that are actually allowing them when they think they actually have something that people want to be able to get way further on the way and so you know I would of course then be like, "Oh, you should, you know, you should learn from this person. You should learn from that person." But I do think that it’s the underlying mechanics of how to build effective organizations that sometimes like we force our nonprofit leaders and social entrepreneurs to just figure it out rather than saying actually like here are your peers or people that you should go learn from who are doing this you know like at this level and how and here’s how they did it. Like so you have the mental you know it’s easier to do things if you have the right mental model right so I I remember asking Dean Cayman okay so first robotics has 600,000 students in it every year I was like Dean like how did you get it to that like you know there aren’t that many programs that have that many kids participating every year right so I asked the team like what is there and they have they had like all this rich information about like well we use parent advocates the parent advocates call school districts that don’t have a first program and say why don’t you have a first program and they’re like that strategy ends up mattering way more than all these other strategies I was like oh that’s fascinating you know but so I that’s the part I find really interesting which is that it is that the field actually knows more than individually than it knows collectively and we can make more progress if we systematically Think about what is it that our best folks, our best performers in a field are doing and how do we actually help more people who have sort of level up on that rather than you know most you go to an education conference most of them people want to talk about the problem right you know we just had terrible nape scores come out people want to talk about that I you know like we should talk about the problems but I actually think like the the conference that often needs to happen is like you like how do I build highly effective scalable organizations against that problem and then it’s like you know well you know you know like do more you know and that’s like that doesn’t like give you a real road map yeah we just I mean we just finished finished conducting a part of new school’s venture fund paid for us to do like a pricing sort of like a I wouldn’t call it a competitive analysis more like a pricing model.

So we talked to again maybe co-pitors, peers to figure out like it’s actually very similar. So Emma nearing our target from achievement network which I would say is another sort of like reference nonprofit that you know massively scaled had a similar model right is like you you’re providing services to school districts like very relationship relationship sales sort of co-unded through philanthropy but there has to be sort of like a skin in the game on the part of the district. But I mean even in that pricing analysis you know most of the peers that we talked to were kind of just like yeah we’re basically winging it you know there were no oh well obviously you know like this is sort of like the expected you know value of a certain you know amount of training and for this amount of PD for and which is kind of surprising given that you know what we were looking at is like professional development teacher coaching you know advisory services these are not you know niche offerings to schools every single school in the country is paying for these things.

What they pay is actually often almost all over the map. Sometimes it’s grant funded and it’s sort of like schools don’t even see the cost. And so there the the there is like sort of weird pricey elastic or like lack of elasticity in terms of you know price and and and the fact that we’ve gone through so many accelerators and even in throughout those accelerators you know nobody really had answers to these questions.

It was often as kind of kind of as you said it wasn’t that there and I don’t think it’s because they they lacked the the expertise or the smarts. It was just there like this stuff you kind of have to do it completely yourself. You have to go out and sort of like hustle and have the conversations.

I want to I want I want to talk about you know one of the things you said is you know a key ingredient to sort of deploying capital effectively especially and I’m making an assumption here that part of what is driving the need for more let’s say deep expertise in the part of philanthropy is you know many of the big problems that are folks are trying to solve also involve you know increasingly cutting edge frontier science. Yes. Which is moving faster and faster.

And so you have this and I don’t I don’t know exactly what the curve would look like but you know one can imagine basically starting in like the 80s 90s suddenly sort of just the pace of you know what’s coming when you started at the white house in 2009 I guess Facebook was just starting had just gone sort of broad I remember I signed up for Facebook in ’ 07 Twitter launched in the fir in President Obama’s first term I don’t think it was even I know Twitter impacted the campaign through yeah it was omni video but that was that was like it. So I mean even even in your time in the White House, I mean there was like all of these technologies that we now sort of look around and almost define society that we live in today, the media infra, you know, the media infrastructure, the media ecosystem rather. And it makes a lot of sense, right, that fund that foundations and funders need I think they struggle to keep up.

And so I guess my the question that maybe we can workshop together for a bit is I’ve talked to a lot of foundations now done a lot of briefings professional learning you know sort of different I mean I don’t I don’t report to go and sort of like tell funders what they should be doing. This is more about like let me let me try to help you fill a specific gap that you’ve identified which is you have all these foundations the ones that we talk to are in the education space. They’ve built deep vertical expertise in education and often within like specific components of that like early childhood or workforce or economic equity social justice social emotional learning right and there’s now this horizontal technology and I think for 20 let’s let’s call it like you know 22 through like almost the end of last year I think there was this begrudging un like admission that you know AI I plus education is not a separate vertical that we can just sort of opt out of, right?

I think most funders when you talk to them, they’re they’re realizing, okay, this is kind of like the internet. This is something that or computers like this impacts almost everything that we do in some way. I I don’t I think most funders will admit that they don’t have the the the expertise internally.

And and I’m curious what the solution is because I mean on the one hand surely all funders should invest more right in capacity building for their teams right AI while it’s very complex you don’t need to be a deep deep expert in artificial intelligence to start making having smarter conversations and talking to the right people you almost need someone like a version of me right like I’m not a deep deep expert in AI I know exactly who I should talk to answer any question that would come up right that took like 8 to 10 years for me to build that. That’s not something you can just sort of like figure out and wing and I think your experience at the White House kind of similar. But the challenge is I don’t think that we you know the the the pace of foundations you know this is a very slowmoving ship that turns you know it’s a tanker ship that that turns very slowly and we need to figure out how do we how do we like sort of adjust tac for this moment and at the speed that that’s required.

I think it’s a great question. I think first let’s just put the like the problem and then we can think about the solution. I think the problem is that it’s a it’s a people flow problem is the way to think about it which is that and the people flow problem has a couple of different levels.

One is every se every sector has its own biases about who they consider like people who can speak with authority about that sector. So in biology there’s an agism bias. So there’s like if you’re going to say smart things about the future of biology, you should have a PhD and you should be you know have had a lab and be this far along in your career.

So there’s like a, you know, like one of the things you will often learn when you get into biology is like there are people who are 30 who have incredible ideas, but they’re like, "No one takes me seriously." Right? They’re like growing out their facial hair. Like there just it’s just like a you have to check a lot of boxes before people say, "Okay, like you’re in, you know, like I want to hear what you what your idea is." Computer science doesn’t have that bias.

Computer science has like a young guns bias. I was going to say there’s a different bias definitely, right? But it like it doesn’t have that bias.

Computer science has a bias for generalizability. Everybody in computer science wants to build something that does everything. And if you said, well, do you want to specialize and actually think about what this computer science application means in weather?

They’re like, that’s someone else’s job. I want to build the world’s baddest, most amazing model that will solve everything. Right?

Computer science has a it’s like you’re trained to be like go as wide as possible. Build the thing that does the widest amount of things because you want to be the at the bottom of the stack of the widest capability curve possible. That’s the instincts of the sector.

So educ so they’re looking for people who kind of have the because these are young guns but so ultimately if you if someone’s young what you’re actually indexing on is ambition. Well, yes, but I’m just I guess what I’m saying is every sector has the things that they point to, but then they also that also creates the blind spot. So often computer science the it tilts towards platform, right?

I want to build the thing that everyone else will use. They will figure out how to use it. I just want to make it.

So this is what’s happening with the models. Like we just need to make the models more and more powerful. Oh, do you have anybody on your team that’s building out the applied use case?

Well, that’s like that’s for the field. Like we will build it, they will use it, right? Similarly in education often the bias is like school systems are complicated.

Learning and school and education is hard. So to really do it well, you need people who are deep experts about the institutions themselves. Like you’re you’ve taught, you’ve been in school systems, you’re an expert about how the systems work.

So for lots of very normal reasons if you are a funer who funds into education you staff yourself with education experts people spent a lot of time in the sector. The the other side of that coin is that they are not like native to the technology. They’re not coming out of the tech sector or out of you know cutting edge research labs with like kind of an intuitive sense of where the techn is going.

Mhm. They’re like you know like you tell me you know they’re like educate me about the technology and I will then develop a first principles view on it. So what that means is that so that’s one challenge which is every sector has its like what it has its biases.

The second problem is that as you said when things move fast there’s always like how do you actually stay on the like where the capability is versus like you’re saying something that is just like a refactored version of other people’s smart thinking. You like you don’t know it well enough to actually say like where things are when things move slowly. Like a lot of people actually have pretty good ground truth expertise about where things are.

When things are moving fast, the people who actually have an understanding of where things are and where things are going get small. That’s I think a second problem like in a fastmoving area. But then a third problem is that when something is moving fast it’s illeible, right?

So I often find that if I talk to technical people who are smart about a particular space, they will have intuitions around, you know, like what would be really useful is if we had a tool that did this and you’re like, well, why is that tool useful? And they’re like, well, there’s just like a lot of interest in going in that direction. And right now it’s hard.

So if that tool and if you tried to like write it up as a proposal, it would just sound arcane. But they have a like so you know there’s a legibility bias. You have to be able to explain your idea to people who are not in your domain.

A lot of edge ideas don’t do well. They they make sense to the person. They feel it like you know.

So like I was telling you about these amazing program officers. The ones who are really good like sense that a space is ripe. If you actually had them try to explain it to you, they would kind of but it wouldn’t like sound like the most compelling pitch.

They just haven’t because they’re keeping up with the technology. They have a technical intuition that pushing in this area is like a promising area to go. You put all these three things together, it it means that if you don’t have good people flow, you don’t bring you don’t have a way of bringing people who are on the edge into the decision-m process of how you deploy money, you’re always kind of you always have the legibility gap, which is they have to explain to you why it makes sense.

You have all your biases for why what is the right speech act? What do you have to so show me? And so, you know, I have this happen every year.

We run this annual tools competition, the learning engineering tools competition. So, I am on the judging panels. So, I watch we we have lots of funders who judge.

And you can just tell there’s like this interesting sense where we have a bunch of technical people who join those review panels and then we have, you know, funders who join and and you will the technical people will be like, no, that’s that project is super interesting. And then somebody else will say, "But how is it going to matter to teachers?" And they’ll be like, "I don’t know, but it’s super interesting." Like, if they can keep extending that voice technology, like that could become the next big standard, right? And so it’s like they’re talking past each other cuz of course you need to eventually figure out how it matters in schools, but that doesn’t necessarily mean that it’s not the technical next edge is not true.

Like that might actually be a really useful So, one of the reasons I’m attracted to this funds model is that it solves the people flow problem. It’s like what DARPA does, right? Why does DARPA say the day you walk in on your badge is printed your last day, right?

So, a DARPA program officer is a 5-year sprint. You bring in, you pitch your program, you get hired to run your program, you have your last day listed on your first day. So leaving is part is the point right you’re here you have the edge expertise you’re going to rush to deploy that edge expertise against the thesis right and then you’re going to go off and do something else which means like you don’t have to have like an awkward conversation around I know you came in because you had that crazy idea about we’re going to solve we we’re going to build you know like when Dan Watandorf pitches DARPA in 2011 the idea of hey my program idea is what if there was a global pandemic and we did not have 12 years to design a vaccine what would we do current vaccine technology takes 12 years what if the world shut down I want to be able to develop a vaccine in 6 to 12 months that was his pitch that was Dan Watandor’s pitch in 2011 was like that’s an interesting pitch like we wouldn’t like how would the military function if people we were too scared of letting them go outside this is a military need like go work on it.

But then when Dan like finished that program and handed it off, right? And which became, you know, like a bunch of the mRNA technology that we then had, he went off and did something else, right? He didn’t say, you know, I should become an expert in AI and like stick around and try, right?

He he’s got his focus area. He had his clear conviction. He made his mark and then he went on.

Right? If you can build institutions that have good people flow, people come in, they’re on the technical edge, they deploy, you just give yourself more shots on goal because otherwise you have the risk which sometimes happens where you hired this person because they were an expert in this thing, you know, then like the organization said, you know, we also need to get smart about AI and everyone is like kind of like reading up on like AI textbooks try to get up to speed like you’ll you know you can definitely get smarter on it but are you going to have the same technical edge in you know being able to come up with really interesting ideas to bet on that someone who like like is obsessively been working on it and sort of sits at that and so my view is that it’s not an eitheror but when you’re doing stuff at the on the technology side on the science technology side. Giving the ball the actual technical expert and saying what would you bet on like actually changes what gets bet on because they bet on things that are a little bit hard to describe, right?

So, I mean what happened to Dan which is like like a clear example. So, he pitches this idea that we we should have fast vaccines. He has a bunch he has a bunch of different targets.

One of them is mRNA. Ebola happens in 2015. So I missed that.

So mRNA actually was used to develop the Ebola vaccine. Well, so here’s the story. So he’s work he’s been working on it, right?

Sitting in DARPA. Ebola happens. So I’m sitting in the office.

This like you know Tom Cle who’s the deputy director gets this like you know he’s like oh Ebola is happening like there’s a supplemental package moving through Congress around R&D for Ebola. So Tom knew Dan. So Tom calls into DARPA to say, "Are you working on anything relevant?" DARPA says, "Actually, we have a program." And so Tom talks to Dan.

He’s like, "Oh, this is interesting. MRNA." So he sends that idea into the policy process for what should go into the supplemental. And everyone says, you know, all the other science agencies, and you know, Tom has this joke where he’s like, you know, if they ever do a FOYA, they can see, they all say no. They all say like the problem is it’s like too speculative and Ebola is raging right now.

Who knows? Like it it probably won’t matter to this crisis. Like we shouldn’t do it.

So they like work over the weekend and convince everybody so eventually does make it into the supplemental. So mRNA gets funded. Now on one level they were correct, right?

It the mRNA technology did not help us resolve the Ebola crisis. We figured out you know it didn’t like didn’t run for many years like within 8 months they were able to just put it down but that investment then ends up making a huge difference in in 2020 right so my like the intuition of edge experts you know that are like there’s something here we should push is a powerful thing even when you know like at any given moment you you’re going to have people going to be Well, is this ready to deploy? Like, is this really going to solve the problem right now?

And like, it might not, but you know, it’s just a question of time scales. That same bet then pays off in this like massive way a few years later. So, the question is is like who’s right, right?

The they were right that it didn’t pay off for Ebola, but Dan was right that it was it was a it was an important portfolio bet. And so that’s the part that I always on your question around you know like how do we actually keep people betting in science and tech I think like the biggest advice I give donors is that this is not about learning this is about recruiting. Yes.

Well, I mean, I I think my my modification to what you’re saying is I don’t think I don’t think that what you’re describing is actually an answer to how does institutional philanthropy sort of adjust tech because you know it to me this makes sense like and I and I think you’ve gotten I think Walton was one of the funders that was part of the big announce that you made. Yeah. Walton does a lot of institutional philanthropy.

And it’s not that they have changed the course and said, "Oh, we’re only going to invest now in deep tech." it does make sense to me that that philanthropy should have, you know, at least some of your capital. You want to like you should be spreading it out in the same way that right that investors do. Yeah.

So, it’s a it’s a portfolio bet for them, right? They’re doing a bunch of other things. I guess all I’m saying is at a portfolio level, it’s useful to almost think of it as like you’re betting on not just a theme.

Yes. You’re actually betting on technical experts with views and that they sort of push ahead, I guess. Yeah.

So, 100% agree. I I’m interested as well is beyond sort of betting on the technical experts who are, you know, in that part of the portfolio. I think the other challenge that I’m seeing is you know in the year following I’m gonna I’m gonna talk about chat GBT is sort of like the anal like the standin for basically all LLMs most of the time when people come to us and they’re they just use chat GBT right to talk about language models generative AI and and this isn’t just funders almost any organization especially the big ones they’ll come to us and say oh can you do you know training for our staff can you talk to us we want to we want to begin the process of sort of like leveling up and and that they generally will say, "Oh, here’s what we want.

We want you to show us how we can use prompt engineering to to do grant making or to do XYZ." because there’s this assumption that the challenge they need to solve is like how do we implement this specific technology right now? And I I mean I I I’ve been invited back enough times that I think even even when I ignore that request, it’s it’s it’s gone over well. I almost always ignore the request and I go in I say I really want to zoom out and sort of talk about sort of the arc the arc that we’re on because really what matters here is not how to prompt engineer and to do grant review and and candidly I don’t even know that I would replace the review process with too much LMS I mean that I think it could be helpful we could talk about that but that’s actually really almost a sideeshow what you what you should be thinking about is we are now on a very high velocity trajectory and you need to basically answer the question for yourself or try to answer the question, you know, are we on a straight line, you know, trajectory is there’s a lot of reasons why we why we might not be.

But you can’t just hope that things slow down and that this is a 10 or 15 year problem because it very well could be a 5year problem. And and when they sort of see that and and what I usually do is I go through sort of a lot of the tentpole applications of not just language models but really sort of this convergence of LLMs, genai and like sort of robotics, machine learning where you see like okay you know now we have you know u Sora is an great example of this where you have this emerging capability of simulating physical environments and it’s like and now and within a matter of weeks there were labs showing, oh, we would just now, you know, first shot have these quadriped robots that are balancing on an exercise ball. Like never like that was that that would take a year in in a lab, right?

Training this robot. And so it’s like, well, this clearly doesn’t have anything to do with education, you know, on the face of it. But what this really means is that all this other stuff that is sort of now bottlenecked by sort of the lack of you know dexterity of different robotics it’s pretty plausible that that’s going to start to accelerate and what that means is the the jobs landscape.

This whole this whole thing about oh Gen AI is you know the knowledge workers the ones that really have to worry about this. I don’t know I don’t even know if I agree with that because actually you know maybe yes over the next three or four years right maybe specifically coders I don’t even know if I’d say old knowledge workers but if we’re talking 10 which is really the the time scale if you’re in middle school you’re thinking about a 10-year time scale as your first job potentially 10 years away you know a lot of the lowskilled you know high complexity service jobs start to become a lot more you know plausibly replaceable And I guess the challenge that I have is I you know AID is not we’re not a solve for all of this like this there is really no one organization one technology one approach that’s going to address this this is actually like we really need every single institution to be starting to map out like what is about what is going to happen right and I think what’s also hard is and I think this does dovetail with what you’re talking about in terms of like where do we find these centers not necessarily of just expertise to tell us what’s happening. But it’s like the smartest people will not answer the question of what are the jobs of the future because the smartest people know that we really can’t predict exactly what those jobs are, right?

We can we can try to take a best guess and say oh it’s going to be you know things where there’s you know that’s a human communication and empathy and creativity but you know I’ve talked to folks who are like actually that’s also right on the table and that’s not that far off that you know AI could even start to you know augment or even replace you know some of those sort of durable or soft skills and so I guess the question is in addition to sort of like the frontier work in science how do institutions that are doing the work in schools and at the systems level start to build that intuitive ability so that they are you know making smart decisions and I think you know AI readiness AI is a great example of this because there’s a lot of bets now being made that are very focused on prompt engineering focused on AI literacy and we’ve now been pushing very hard to say it’s not really AI literacy it’s like mobile phone literacy sounds ridiculous like Google search right literacy like nobody there is no class where you learn these skills. They’re they’re ubiquitous and seamless. And we should expect I mean I think it’s a very reasonable assumption to make that AI is sort of following that path.

And so sure AI literacy is a component. We teach kids to use Google search and we probably talk about their phones less than we should. We probably need to do a little bit more education in terms of like digital you know digital citizenship citizenship and you know online safety.

But there’s there’s a lot more to what how schools need to evolve that is touched by AI but not necessarily a a problem of how do we implement AI and so so it’s like that’s a very hard and complex conversation to have and so often times I’ll get into it with you know with different organizations and you know they might be very compelled by it but they’re just like they don’t have the cap capability to now go back to their stakeholders and sort of translate Right. And you just talked about sort of that same a very similar sort of like challenge on sort of the deep tech side, but I feel like there’s the the organizational like to me the big challenge of AI right now everybody’s like obsessed with deepseek. And I do want to talk about that a little bit if we have have a chance.

A lot of the experts that I’ve talked to have said I mean yes this is a big deal for stranded capital and sort of like these big models and we have to figure out if like how cheap it actually was and is this really a paradigm shift in the training. But even if it was like radically cheaper, the question was like is this going to accelerate implementation in education? Like disconnect right and most of the people I talked to said probably not because the affordability of models has not really been the li limiting factor like organizations have been struggling because these are huge bureaucracies that just don’t have the muscle memory or the capability to like adapt their workforce capabilities and really think about what does it look like to sort of adapt or retrain or sort of like adjust hierarchies or and yeah I think I think it’s two different things.

So I think one so when I joined back in 2009 so part of the reason the the first first portfolio I got was President Obama had just given a speech at the National Academy of Sciences. This is coming in from the Bush administration. The big theme of that speech is putting science in its rightful place.

But then the president put half of that speech was actually about math and science education. The president had this line which is we need more engineers and less financial engineers, financial crisis, right? So part of his sort of view was that like what is the economy we’re trying to build and what is it that our children deserve to be ready for that economy that’s going and this is in the midst of a financial crisis where part of the big conversation was are we too financialized you know coming out of the great financial crisis there was a lot of question around what is this you know I mean we it was tough right like what is this what are we building towards but what was interesting was we did I took that portfolio kind of like STEM as an acronym was not really used, right?

It was like kind of an obscure academic idea, but like science, technology, and your math, we would always have the president say it out loud rather than use the acronym. But we started to work on it, but the at the time we did some polling that that we came across which basically boiled down to a headline, important but not for me. They interviewed parents and they said, "What do you think about like your kid getting trained on this stuff?

You know, getting access to advanced science courses and math courses and, you know, all this stuff." And they said, "Well, you know, my kid’s more of a reading kid. My kid’s more of a writing kid, more of an English kid." This kind of thing. It’s like important.

I think it’s important, but maybe it’s not right. So that’s 2009. You basically fast forward three years.

So 2011 2012 Gallup does a survey and Google gave them some support a couple of different people but Gallup does a survey and basically starts to ask do you think that it’s okay that only 25% let you know somewhere between 10 to 25% of high schools offer computer science like would you want computer science offer your kids school 90% of parents say I want it offered at my kids school there’s no difference between African-American parents versus Hispanic parent, right? So, suddenly you’re starting to see this like cultural shift like I started to just bump into STEM camps, STEM programs at schools. It’s like so we went through this like cultural shift.

The number of students who then get to college and express interest in computer science as a major like first quadruples and then goes up 10x, right? You’re starting to have schools that don’t offer computer science. It’s maxed out.

We go from economics and psychology as the baseline most common degree to CS this all right the interesting thing is like we somewhat like won the culture war right which is like there’s a huge surge of demand by by students and by parents but then the question was like did the system actually then right this is way before the most recent wave on the on the AI side parents were already like like this is kind of the future. I want my kids to have access to it. You know, students wanted to get access to it, but like the system hasn’t really caught up, right?

We still have shortages at the college level for people being able to get a computer science undergraduate. Right. Right.

And that’s like criminal, right? How is it that you could enroll somebody in a university, they want to be a computer science major, and you’re like, "Oh, we just don’t have enough faculty to let you be a computer science major. You have to pick some other major." Like, that’s the future, right?

So, I think that there’s there’s ongoing work that we need to do to explain to people how important it is. But I think parents are already there. I think like a lot of students are already there.

They want they want to be there. That the question is is like are the institutions actually making the investments so that if you want to be able to learn about this stuff you want to be actually not just be a consumer of it but be a maker of it you want to be able to keep progressing right so one of the things I always say is like why don’t we have an AP engineering right so like yeah why don’t right like we haven’t actually added STEM APs in all these things that we value for a long time right we have like an art history AP. We don’t have an engineering AP.

So like if you wanted to say like I’m progressing and I want to show that I’m actually doing cutting edge engineering work. What are the summitive courses that exist in high school and have any have we created any flagship new ones in the past 10 years? And there there’s also even lowerhanging fruit you know Zera data science for everyone and I mean what he’s described when he describes you know his vision of not just like a course called data science in every school but like just providing data sets so that social studies teachers or English teachers can have kids go through you know he was like oh you we can give kids access to like you know Spotify data sets and I was like I want to that sounds so freaking cool right and just the idea that some of this obvious stuff has And but but I but to me this is a really exciting place where philanthropy should feel incredibly empowered because there is no other mechanism in my view like I think the the private sector I think edtech really burned a lot of bridges where it’s and I think the productization of AI is sort of is going to is definitely going to lead to sort of a repeat heartburn where you know schools right now they’re having like literally thousands of vendors coming in like pitching all these AI tools and and so I think this is where Flameth can actually come in and sort of like think more holistically.

And there’s obviously work that’s being done and yet it seems like my and maybe you can validate this. I was talking to somebody at I don’t know if I should say one of the big education philanthropy consultancies and they were saying that the K-12 philanthropy space is actually starting to get smaller. And their explanation was that a lot of funders who’ve been doing the work for a long time are sort of exasperated at the just the lack of progress and it’s just so complicated and multi-dimensional.

They feel like we just haven’t seen the return on investment. I am curious if AI is sort of this sort of opportunity to reinvigorate that because it’s it is this new shiny object and the question is how do we make sure that it doesn’t become so shiny that it you know computer science comes to mind right like the the there are a lot of folks who are saying well do we even need to teach computer science anymore now that AI is obviously really good at coding and I’m like yes we need more we need more CS we’re not even close to done with that work so I I’ve I’ve always been sort of adjacent to the education community because like I’ve always had my foot in always working on science and technology and and v variety of fields. But one thing that I’ve sort of observed being like adjacent but working in education for 15 years is that I think the field overall is at a you know it’s not it’s at a low confidence point as to what does it strongly believe.

Right? So when I came in to this world back in, you know, 2009, you know, I guess I was the lawsuit we were doing was before that, you know, the the leaders in the field had I mean in some instances potentially incorrect, but strong opinions. They said, you know, we need to have internationally benchmark standards that we are teaching kids to a high standard so they’re actually ready for college and career.

We need to have performance benchmarks for teachers. We need to have high quality assessments that are actually touching you on a range of skills. We need data systems to track.

We need to be closing down low performing schools. Like there was like a a huge like kind of agenda for what the school reform look like, right? And it was controversial, but it was like it had folks working on it.

You know, step by step, the political coalition that spanned left and right for we can do more and do better slowly fell apart, right? It fell apart on standards, then it fell apart on teachers, then it fell apart on assessments, and then it fell. So, it like it just kept losing legs of the stool.

And so part of the question I think is that like there’s not you know people ultimately said well you know it can have many goals and you know you can you can pick which goal but we don’t have like some strong opinion about where we’re trying to take the sector. I think part of my goal for why I I sort of talk a lot about education R&D that one of the roles the federal government can play is actually funding the R&D of it is that in a world in which people don’t have you know we’re in this place where different districts different states everyone is sort of going in different directions one place where you can make a meaningful contribution is at least trying to make the R&D better. But I think this idea of like can we be doing more at a coalitional level to make sure our kids are ready is something that you can do even in this political environment because if you say look we don’t want just our kids to be AI ready we want you know we want them to actually be like feel confident and building with AI well that actually can involve churches that actually can involve a whole bunch of cultural institutions it can involve schools it can involve like it can actually feel all hands- on deck where everyone is like well what is my part I’m building the data sets I’m providing this is the way the afterchool community can be part of it so even in a world in which the it’s not the like we’re going to like do this this one big thing and then it will like education will be on a better pedestal you know you could still build what I sometimes consider these coalitions of the willing and I actually think on this sort of like how will AI integrate and how will we be ready the coalition approach can actually be quite powerful.

So that’s the thing that I think on the optimistic side the the work you do is actually more possible and just as possible in this current political environment. If you told me well you know the only way I make progress on whether kids are ready for AI is if I get this one big important policy win and if we get it we win. If we don’t get it we don’t win.

I’d be like you know I don’t know if that’s going to work out. Yeah, but we I mean our our biggest success story to date is is still the state of Ohio, which people, you know, when I was I was growing up it was a purple state is now red, maybe even deep red. Mhm.

And it it’s our messaging was exactly the same. Like we did not have a separate set of talking points. So, like we literally were talking about and I think what what resonates is, you know, a lot of parts of the country that Republicans you know, associate with these communities that are sort of like left out of this conversation around sort of like equity and sort of like economic access and empowerment are are communities that resonate with the story about, you know, the changing workforce, right?

And in Ohio they had you know they had the industrial revolution and I mean this was sort of mechanistic automation and also like offshoring and and the experience of becoming you know a member of the rust belt is still very visceral to them and so there was an intense desire to say okay how do we get ahead of this and and and the same it’s that that same vantage point is is in Michigan purple bluish state and both states are interesting because they have local control and so the The reason we were so successful in Ohio was also because we were going in and saying we’re not trying like our success for us is not Ohio the governor or the department of education like launching this big initiative. It’s like we’ve what we’re going to our role is we’re going to identify this coalition of community based organizations. And I think what’s and I think what’s what’s maybe unique about maybe not unique about our approach, but I see echoes of our approach in in certain other very what I would consider to be highly successful nonprofits like data science for everyone where our goal is not to get really big and to become the teach for America with thousands of core members in every single state sort of like doing the work and you know success for us is actually staying relatively small.

And that allows us to now go into these communities and say, "We’re not here to just be, you know, we’re not here to ask you for help bringing aiEDU into Youngsttown, Ohio. We’re here to figure out how can you become the like the the organization or that superintendent that brings your community into the future. And we’ll give you the tools.

We’ll help you we’ll help you do that." I I am interested in, you know, I think this model really works. And we just had we were just in with Washington state this week for a series of statewide summits where we were pulling in this is money from Google.org sort of bringing in rural educators. A lot of the AI summits don’t necessarily happen in places that are accessible to rural educators.

So a lot of the work we’ve been doing is like getting them scholarships so they can go and attend, right? You’re building it’s like the actually the Obama campaign neighborhood precinct model. It’s like you build the platform, you give the resources to this person.

You know, you have to pay them if they feel like they are change makers for their community. You end up seeing like incredible amounts of work and drive. And I think that’s been replicated in education.

I’m curious outside of the work we do, there’s like all these other components that you have visibility into in terms of like the cutting edge of science. And but what I like about your approach is it feels messier, but the bottomup approach I think has more durability. So I think that one of the things I one of the things I always encountered when I worked on education was you know I was for example very I was fascinated by the maker movement right this like idea that oh there’s all this interest around making allowing students have the tools of production you know like what’s the 21st century shop class and you know we were like we were like culturally interested in how do we bring back shop class as like a 21st century concept.

So, we hosted the White House Maker Fair. We did a lot of things to sort of celebrate this idea that we should have more dedicated spaces where you do long-term engineering projects that sit that they don’t comfortably sit with art or engineering or, you know, like they sit somewhere in the middle and they allow students to be ambitious in what they’re building. What’s interesting for me is that like whenever I go anywhere, there’ll be like tons of maker spaces.

Like people will talk, oh, do you want to visit our school? And they’ll like show me their maker space. They’ll have like pictures from, you know, sometimes they’ll have pictures from White House Maker Fair there and all this stuff.

And like we were like not setting a lot of policy, but we were celebrating this bottomup movement where people said, you know, these classes feel too rigid. We need to give students the space to set ambitious projects that they own, that they’re going to build, that they’re coached on. So, what I like about your approach is that you’re saying like, look, there’s a lot of people that want to like lead.

They don’t want to be like, oh, Chachi BT or AI is this thing that will happen to me. Like, they want to be part of the story. They want to make sure the students are part of the story.

They want to be part of the story. They want to be part of how this will roll out. And you’re actually giving them like a format, a voice, a structure to do that and saying to them, the political leaders, well, this is actually what leading will feel like, you know.

So, I I guess I mean, obviously you believe this, but I think the reason why I think it’s useful is often this is considered the kind of fuzzy part of education. Like it’s like, oh, that’s the feel-good part of education. Like, oh, we’re doing stuff and getting people together.

And then there’s the serious part of education. And I think what people sometimes misunderstand is the bottomup stuff ends up often being the stuff that you come back 10 years later and it’s the stuff that’s like now in continuing to grow. Whereas lots of stuff that’s like sort of the hot idea just kind of burns off.

You try to you you don’t encounter it you know as you sort of visit schools in the future whereas so I think you know one potential thing is that I always try to remind people is like what has staying power what have been ideas what have been products tools movements that like have a lasting impact on how education works versus not so you talked about the importance of having having folks that that are really in in in in growth in these frontier technologies that you know may not be able to know exactly where we’re going but they have this like very you know potentially powerful intuition. Mhm. As I was going into 2025 and people would ask me sort of like what are your what are you expecting for, you know, AI this year and I was, you know, I think like I I was preparing for this sort of plateau moment where, you know, all these new models start coming out and we’re starting to see incremental but maybe like, you know, decreasingly interesting capabilities and sort of we sort of at at this place where it’s like, oh, a lot of the the cynics who were telling us that at some point we’re going to hit these walls where you know we’ve already run out of you know essentially high quality data synthetic data is polluting the models and sort of basically useless and you know either the more chips won’t actually do anything we’ll sort of like there’s no rule that so far the scaling law said that like the more GPUs you throw at AI the better the the models will be but there’s we don’t necessarily know that that will continue sort of indefinitely or maybe the chips just get really hard to to to build and And then Deepc comes out and and also you know 01 and 03 from OpenAI that you know are and I know benchmarks are kind of fuzzy but we’re seeing we are seeing increasingly interesting capabilities.

It really does not feel like we’re slowing down and and one of the biggest challenges I guess the other hurdle that I didn’t mention is cost. You know if these models remain wildly expensive or energy hungry then they will just never be useful enough at scale. So so now we have this model that you know perhaps to some degree is more efficient.

And whether or not that approach itself is going to be the game changer it it it does feel like the all the people in the AI community that I talked to they’re they’re they’re talking a lot more seriously about the the like like there is this increasing confidence that we’re on this trajectory, right? So my so this is a two-part question. One like where are you on that?

Like are we sort of is this sort of are we accelerating decelerating or sort of just steady state which has been quite fast obviously from 22. And then however far out you’re willing to sort of portend you know what’s in store over the next not and I don’t want to go too far out but like let’s say 5 years like a me like a meaningful amount of time but but near-term enough that it’s like worth people paying attention to and actually preparing for. Yeah, I think one way to think about AI is to compare AI to crypto, which is one thing I always say to people is what is the what is the delta between what how much public enthusiasm there is about something versus what do technical experts think?

Like how like if people who like kind of understand the core tech, do they think it’s a big deal versus like kind of just broader enthusiasm of people who are trying to monetize the technology? You know, I would say like crypto has always been something where the people who actually understand it technically are like this is good technology like a base, you know, having a ledger technology that is like the blockchain maybe more specifically, right? Yeah.

Is like powerful, but they’re like, you know, like a it’s ledger technology. It’s like useful. But there’s like, you know, it’s not like they when they they’re not spending all their time being like, "Oh my god, things are happening that are blowing their mind." AI is an area where when you talk to technical experts who follow it closely, they are more like worked up about the pace of change than people who are watching from afar.

So it is like a it’s it’s inverted. So I think one of the things that is not that I’m like in the prediction game, but I think one one sort of mental model you should have is like what do people who are closest to a technology on the technical level think? Do they think like oh this thing is happening slower than we expected?

That’s interesting. Is happening faster than we expected. People who are close up to it are startled by the pace, including people who aren’t incentivized to fuel the hype.

Exactly. It’s not It’s not just the company speaking in their corporate voice. People who have who are, you know, in settings where they’re doing it on a personal level.

They’re they’ve left these companies. They’re tracking, you know, a lot of different sort of and and and the researchers that have been in this field for like in some cases like 20 to 50 years. I was just talking to Peter Norvik like who like like he was like I’m like the the the AI hipster.

Right. Right. And you know I mean it’s I think he’s a little bit more circumspect than others.

He’s not you know I don’t know if elated or sort of like sort of brimming with excitement is how I would describe Peter. But he’s definitely not skeptic you know I think like folks like that are who I would turn to to kind of get a sense of like is this are we on the hype train you know so I think so I think one thing is just that technical folks are genuinely think that this is something important and progressing quite quickly now how quickly it’s open question but you know I think that’s I think one thing I think the second thing is that I think it’s sometimes hard for folks who are not watching it closely to keep track of the pace of change. So one you know data point I sometimes give people is just like you know take something simple like cost for a million tokens where you know maybe 18 months ago it’s like $36 for a million tokens you know today that might be 18 cents.

So, you know, that’s a, you know, 500x, right, sort of improvement where you you have a reduction in cost. That is like quite substantial for an 18-month period. And you know, like new things are coming out all the time like Deep Seek that might be like that might knock it another on the on the cost curve.

So the the way I try to think about it is at least in the short term like next couple years we are living through what one could describe as a massive capability overhang which is for lots of like economic reasons and the way the field is structured everyone is racing to build more capability. Let’s make the models better. Let’s deploy more compute to keep advancing the state-of-the-art.

If you then just like look at the other side of the coin for how much investment, talent and everything else is going to exploit the models as in to use this capability curve for real world use cases, education, health, you know, social uses like can we detect wildfire out of the sky by pairing these models with satellites. It is minuscule. So like somebody was describing to me this project that basically you know given the potential of Starship would launch a set of micro satellites pair them with advanced AI models and basically try to catch the signature while of fire around the world with a goal that you could detect a new fire start within 5 minutes of anywhere in the world.

I was like how much would that cost? They’re like, well, you know, if launch costs are where they are and cost of AI models, all in 40 million, right? So, like 40 million on the on the capability side wouldn’t like get anybody out of bed.

You can like they’re like, oh, well, to train one of these models in the open source or these, you know, like 40 million is like the latest models are now they’re nearing a billion, right, for something. I know the GT4 was like 100 million, right? So there there’s an expectation that like you know if if it was 100 million that like this next model if they just the scaling costs continue would end up being in those ranges.

Certainly the million right the compute costs of like the you know the cluster that Elon’s building you know it’s like just at the compute cost it’s a lot then you can sort of impute it towards whatever model they’re building. But the point just being is the idea that you could, you know, detect wildfires within minutes. The all-in cost is 40 million and people are like, "Oh, how do we put the funding together and how do we sort of go after it?" So, I think we’re living in an era where one piece of this is racing ahead while everything else is like still in the like slow lane where it’s hard to find the money.

It’s hard to find the talent. You have to find talent that can bridge those divides and we’re like slowly coming across. So I think if you were to look at that I would say like there’s obviously if you have something you can contribute to improving the capability of the models go at it but there is a lot of gold in the hills around the exploit side of the curve because the the amount of money and the amount of talent that’s chasing that is still quite limited compared to its potential and you get to you get to be the beneficiary of the capability curve.

So that’s the like you know when I sort of look at the space looking back I say well you know are they spending the same hundred million on building the next generation biio model they’re not right so that’s where the next value if once you start to see that then you’ll start to say oh you’re now starting to see the pathway for how this core capability is going to spread into the economy. I mean that was perfect and I think it it’s the it leads me to my last question and and before I we get into that I just want to thank you so much for of course coming on and nerding out with me. This is I learned a lot.

I did not I didn’t realize maker spaces were originated in the Obama administration. Well or the maker fair. The White House Maker Fair.

Yeah, cuz it well it was I mean I think it was like a brain predated concept. Yeah, I was a brainchild of this guy Dale Doherty who used to write this publication called Make Magazine. And there was like this mini cultural movement that started around MakerFairs which were these like kind of hobbyists that you’re like making stuff in your backyard and then you would come to a maker fair and you would show other people what you had made.

But was interesting was kids started to show up with the stuff they were making. And so this idea started to bubble up was this idea that like you know it’s like the old world’s fair like why not just show up with like you know like if you think about the early days of the computer industry you know where does Steve Jobs and W show up to show their first Apple computer they show up at the homebrew computer club. What is the homebrew computer club?

It is a hobbyist club where people who used to build their own PCs, you know, they would get a panel, they would get the keyboard and they would like build their own personal computer would show off their own designs, right? So, one of the ideas that you always have to remember is like hobbyist communities, especially hobbyist technical communities, sometimes just frontr run the next technical breakthrough. Like look to where the hobbyists are.

And so one of the ideas behind the maker movement was is like well you know like there’s learning going on here like we should think about this as an educational movement just like anything else. So I found that very interesting, you know, but like we used to do stuff with it, but people used to be like, "How does that fit like education like policy?" And I would say, "Well, this is actually what makes America, right? The backyard, the garage, the builder, like why not make that into something every kid can see?" Yeah.

And I see I see parallels with we just had Isabelle Howan from the learning accelerator at Stanford and you know they’ve had amazing success with these hackathons where people no background in you know software engineering are showing up and like building these incredible tools. So so for people who are you know they’re they’re bought in now and they’re like okay I want to identify you know how I can start to harness the fruits of all this sort of like frontier you know AI development and identify like the use cases and sort of play this role of the sort of like ambassador of bringing AI to sort of like these you know big maybe institutional you know components or sort of corners of our economy and society. What are how do they stay current?

I mean and and maybe be as specific as possible. Like are there any specific researchers or blogs or podcasts that you can share? We’ll drop the links in.

But yeah, where do you redigitate to stay up to date? A great question. I mean, there’s a couple of different things I’ll point to.

So we actually run a list sererve called the learning engineering list sererve which I’ll send you a link but if folks I mean it’s it’s relatively nerdy for folks who want to keep up on what are the race you know what is the kind of AI and education community up to on technical papers new tools who’s building their next you know company and everything else is an interesting way to sort of be involved so if you’re like a PhD student or you’re coming up in this space and you’re early and you’re like what where What’s the water cooler for people who are kind of building at the intersection of AI and education? That’s that’s I think one. I think you know we run every year a learning engineering tools competition where you know we say at least at the entry track it can be an idea on a napkin right so we have teachers and students and others apply with ideas.

So the next round of that will open this fall. So that might be an interesting one and we can I can include a link to that. I think the other place is that you know there are I think the place where if you’re motivated on this you can get to being on the bleeding edge really quickly is pick some domain you really care about you really care about early learning you really care about dyslexia you really care about some very you know you really care about like some very precise problem in education if you can think really hard about unsolved problems in that sub area that are prediction problems, things that like nobody gets to because we just don’t have the people to do it.

And try to think about what is the what is the intersection between the thing I care about a lot and AI? Who’s doing work on it? What would be the to-do list?

Mhm. I actually think you can get from like hey I’m like outside this space to I’m like actually one of the three people thinking about the intersection between this between AI and you know this particular you know sub area of education or sub area of science like faster than people think because the people who are cross domain experts are actually shockingly small. So, like if you said, "Oh, who’s doing the best work on road safety and AI?" It’s actually like not that many people you would think.

Or using satellite imagery and AI to spot wildfires. That’s actually, you know, you could fit them in this room like and they’re all busy and don’t necessarily have, you know, the and they have ideas that you can talk and you can actually so like people often want to, you know, advance to like I want to figure out where there’s like the open space, you know, like pick something you care about and be like, I’m going to try to figure out who’s actually doing stuff at that intersection. Yeah.

I mean I feel like in in accessibility it’s almost like you know real time transcription and translation it’s like the capabilities are now quite incredible and you know we were working with a couple organizations dyslexic edge and cast and you know it’s it’s still very early days right and I was actually a friend of mine was was asking me like oh is there a tool where you can you know basically use AI to sort of like speak to someone who’s deaf or hearing impaired and it was sort of like use generative AI to create the signs. There’s I maybe if someone can comment if they actually are aware of one. I I was surprised at like how how little I could find online even though it’s just such a straightforward use case and it probably wouldn’t take very much money or very much right you know a big a very big team to figure this out.

But I think in part because the capabilities are are literally, you know, 18 months old, maybe even less, right? For something that’s like truly like usable enough that, you know, you could actually go and put it out into the wild. Well, there’s a great way to end it.

There’s a lot more that we could discuss, but we’ll have to save that for another time. Yeah. Yeah.

Thank you again for for joining and good luck on Renaissance and all of the stuff that you’re hopefully going to be successful in building and funding. Yeah, thank you so much. Thanks for having me.