From Podcasts to Platforms: Grace Gong on Building Community and AI’s Next Wave

There's two major shifts since the ChatGPT era.

The big value out of AI is the ability to do personalization at scale.

There's no big tech company competing with you, then you're not building something super substantial.

Don't expect the AI to be perfect.

Almost every step domain expert has like an AI agent for it.

It'll significantly outperform like ChatGPT.

Not everything could be fully done by AI.

Hey everyone, this is Grace Gong, a good friend from San Francisco and she is joining us today on the Venture Capital podcast. She is like a premier VC podcaster. She's got two shows, one is Venture with Grace and another is called the Smart Venture Podcast. She's also an investor in San Francisco and she just held a phenomenal AI conference in San Francisco called the Smart AI Conference. I flew down for it. I was impressed. She announced it and within two or three weeks had phenomenal conference. I think one of my favorite presenters was one of the CPO for Cisco that talked and like man, that guy, he could deliver a speech and it's amazing.

He's a goat.

Thousands of people there presenting, sharing thoughts. She had an after hour VIP dinner that I broke into like Grace is legit.

Thank you. I'm crying. And Kai also have like this video so I can like clip it to put on my TikTok for sure. Yeah, just to improve my legitimacy just in case. But thank you so much. John, I'm so glad you came to the Bay Area and yeah, I'm so glad you had a good experience at our conference.

It was phenomenal. Why did you start this? Why did you launch the Smart AI Conference?

I genuinely feel like there's like a few things maybe I could start with like the podcast. Like I feel like I know that this was like not necessarily originally planned but like I guess like so I started my podcast because I was originally an international student and when I came to the Bay Area or when I came to the US I actually didn't really have much of a like a business connection. And when I moved to the Bay Area I felt figure like I can just like ran into a lot of these like really amazing executives like on the San Francisco area or like Berkeley or Sanford campus. I thought that have the responsibility. I mean I feel like I'm very mission driven but I feel like I have the responsibility to actually bring that to share with people like my younger self or like people around the world because you know, a lot of the knowledge or like a lot of the things intact or venture capital is very Opaque. So, so I kind of wanted to.

Be.

I want like people who doesn't, doesn't really have the access to actually have access to these kind of knowledge and information. So that's why I started the podcast. And fast forward to today. We have reached like over 110 different countries and on my LinkedIn I got 10 million impressions in 365 days. And I feel like I was very fortunate and I just really wanted to make sure that like, you know, people could be able to access to information that they are not very easily to getting access to. That's why I started the podcast. And then I feel like the conference is kind of like a reflection on that because I feel very fortunate that I've been just doing like a daily podcast in the past like a year and a half. I was able to feature some of like the top venture capitalists or C level executives from different big tech companies as well as like founders who are, who have created phenomenal companies in the unicorn status. And I kind of wanted to bridge the gap between online and offline to find a place for people to all gather together and share their own thoughts and experience. And for the 100 list and the conference, I think about them collectively and separately as I think for collectively because I feel like the offline experience is very important for people who I've interacted online and a lot of them wants to meet each other, they want to find belonging in a group, in a community. So I kind of wanted to create that community gathering experience for them. On the other hand, I feel like the list itself is kind of like a celebration of innovation just because I seriously felt like I've encountered like there's so many lists about like the later stage companies in the world. They are, you know, they're by like the most like famous magazine and everything. But I feel like in the end of the day like it's not the one that who like because like at the stage that they were covering by every press they're already, they probably don't need your help anymore. They probably don't like need the like they already have a lot of coverage. But I kind of like feel like the younger or like the more earlier stage company they could benefit a lot from like being highlighted especially if they're doing something phenomenal. And as a early stage investor slash angel investor myself, I just personally wanted to celebrate the earlier stage company that I've seen come across my desk, come across like my network to help them get their, the highlight that they actually need. And that's Why? I kind of created a list as well.

I thought it was a phenomenal group, amazingly well curated. I think you did a phenomenal job from that event. What bottlenecks in AI company buildings has stood out to the most from you, from the conversations that you've curated?

I think one of the things that I thought was a bottleneck is like how many repetitive customer would people have after this current AI era? I edify there's two major AI era from a business standpoint. One is the ChatGPT era which is the era that everyone discover consumer AI. People started using ChatGPT in their day to day life and people started building on top of ChatGPT to build different AI agents and different AI products to like accelerate like the AI adoption into human society. And then the other part is like the current AI agent era which is I call it the lovable era. So I feel like there's two major shifts since the ChatGPT era. I haven't really, I haven't really talked about this or like read about this on LinkedIn. I should be blogging about it. But I think the major thing is like the difference between like the ChatGPT era to the lovable era is like the lovable era. Everyone and their mom can build their own tool using lovable or like equivalent of tools that they could build their day to day like useful things like a fitness app or something like a diet trucker or like anything in their day to day world. So like the consumer can build their own app and then what does that mean for like the enterprise society is. I think the bottleneck would be like there is no moat in a lot of the technical solutions and you have to build really strong distribution to actually enter a market even for enterprise grade products. I think because of like there is so much of the like there's like a huge group of enterprise was enterprise software are targeting SMBs beyond the bigger enterprise. So I feel like the SMBs are not going to be super careful about like if something is secure and then so I basically I feel like the bottleneck would be you have to build a really strong distribution and go to market. Having a really strong go to market strategy versus if you are just like launching your AI product and then wishing it can just go viral.

So Grizz, I'm curious to dig on this a little bit more because I've been saying the same thing for a while. I think a lot of people have, but I don't know that I've seen a lot of people like vibe code and app, put it out there and start competing like in a meaningful way. I mean there's a lot of YouTube videos of people like oh yeah, I built this thing and like you know that like one of my favorite videos is one where they use cloud code to basically build lovable in like an hour. Um, but you don't see them actually like putting it out there and competing yet, right? Like in a meaningful way. Um, and I think like vibe coding actually to get something that's usable actually takes quite a bit of work and effort. Um, and so I'm curious like when do you think we will see people being able to actually build full fledged applications that they, they can deploy in a way that's like truly democratized? Because I don't think we're quite there yet, but I think everybody can like see it, right? We're like it's almost there, we can almost touch it but like it's not quite right.

I kind of see my disagree just because like I feel like I want to be respectful John John's business because of like you are running a developer business. I feel like this may be like going a little bit against like you.

Can say whatever you want to say. We only have honesty here.

I think like I've heard so like like, I mean I know the names of these companies but I don't think I kind of like public say them yet. But there are one person, $7 million company in the VC society, like you know, in people I've seen that they code something that's when viral. You just didn't know that's a lovable product. They don't really advertise it as a lovable product. Those products are like more towards like the day to day use products. So like for example let's call it like a fitness app or let's call it like a scheduler for a dentist. So like those are not very, they're not publicized as a lovable product. There is a little bit more sophisticated technology behind it. It's not necessarily like I build it within an hour and then it become like viral but I've seen like one person coming or like very few people company that with without like a technical background they just built stuff that's like actually enterprise grade. And then I, I remember like I was having lunch with my mentor and then my mentor told me about like they're, they're like a well known VC firm and then they back like a company that's doing pat related like AI and then like the CEO who is like a non Technical person was like a basically MBA type of people build the app that they're entire company build over a weekend just to demonstrate that like you know, we are not using enough AI tools. So I've seen this is like starting becoming a trend and maybe in Silicon Valley it's more extreme as like people are somewhat technical. But I do, I don't see it I guess like you know, you and I built something that would just sell but like I mean I also see it to a degree. So it just like I don't think people are publicizing as like they are a one person company or like they are, they have built this unlovable because they don't want people to stow certain like ideas identity or like they don't want any massive attack to attack for business because it, it is in the other day not necessarily safe. But I, I've seen that becoming a trend at least in Silicon Valley.

Yeah, well and I feel like, I don't know that incumbents are quite feeling the pain yet from these small like startups. Right. And so there's maybe an element of that too but they likely will be soon.

Yeah, I feel like, yeah by bottleneck I really just felt like it's like because a lot of big tech are trying or big companies in general they're trying AI but I don't know how long they're gonna use a certain product if they don't see a certain result. Because I think even bigger company they have FOMO as well. So they just fear of if they're not using AI, they're losing or if their competitor are using AI. So everyone feel very necessary to add AI to their budget. But on the other hand I feel like a lot of these tools may not have meaningful revenue impact. So like I think those two are going to be cut. So like I think that's going to be a bottleneck. And in terms of like building AI tool that like from a product level actually make like financial sense for people to continuously using it. That's kind of like what I meant by like bot on egg.

Well, I think part of it too is the building in AI is a different flow in a lot of ways than like what a lot of companies have traditionally done. And so like it's. I think your example of that CEO building something over the weekend is a really interesting one because like clearly he has a lot, he or she has a lot of incredible capable people on his team for sure. But they also have all these like inbuilt processes and standards and you know Technical debt and all these other things that make it really hard to move fast. And so I do wonder if like there is an advantage to being just like 1, 2, 3 people that are, you know, building from the ground up in AI and AI native from that perspective. Just that they can move so fast and they don't have, they don't suffer from any like sort of some cast cost fallacies and so forth.

I feel like. Do you mean like if they go fast and then they are like not really like stable, basically the product.

No, what I'm, what I'm saying is like an incumbent like these large enterprises. So there's the MIT study, right. That came out. Everyone's referencing.

Yeah, 95% agents are not out.

Yeah, basically. But also like there's no way like that's actually like true to a certain extent. Right. Like everybody's getting some productivity gains from using some level of AI.

But.

But yeah, like, I just wonder if like these big large enterprises, it's just too hard to use AI as effectively as somebody that's like using it from the get go. You know, it's, it's almost like the analogy of grandma trying to, you know, use the computer and then her grandson is like, you know, dancing circles around her on that kind of thing.

Yeah.

I met with the CIO of a company that has a thousand employees ish today and I told her to use replit. But in their case their org is not built where they can bring all their data and their flows to repl it. But I think it's very possible that someone starting on replit today could working within replit's flows begin to grow. And as replit becomes better, bigger, they can grow and I think you'll see more of that than happen.

I, I feel like it is like for big company to adopt AI. I think the major challenges like I've hear about like, I mean every day I tried with like the founder or VCs. A lot of people are saying you have to do like both like bottoms up and top down to like sell into enterprise as like not necessarily that just adoption, but I guess like, I guess they are essentially a lot slower when it comes to like the security processes. Because I, I'm sure you guys have heard of like these like you know, the airline company that like just like randomly the chatbot reimburse people like just like pay for people's flight for free and stuff. So I feel like it's better that if they move a little bit slower than if they just like enter with like Full force to onboard a bunch of stuff that I feel like it will be not very secure to the company.

Yeah. But they are kind of caught in this dilemma of like if they don't move fast enough, they get disrupted by the up and comer. But if they move too fast, yeah, they're gonna break things. Right. And that's just, that's just reality. I'm curious what other like contrarian or surprising insights did, did you take away from your conference?

I think the contrary. There's like two major takeaways. Like one thing is like the, from the Cisco president, I think he shared that like for any company, if you, if you like basically you have to have like a tier one product then like any. Basically if you don't have like a market pool, your product is not ready. Meaning if you don't like, if you have like a second tier sales team to sell like a first tier product, then your product is okay. So like basically like if you have like a second tier sales team, like you don't. Basically the meaning is like it doesn't take a lot of effort to sell something that's like actually have market pool at that time you have like a successful product company. And I think that's one really interesting take because I've also heard the opposite of the story many times as like people try to convince like you know, 40 buyers or like 40. I guess like I only heard about like the start of that trying to convince like 40 investors like or like 100 investors like their product will work and then Maybe like the 100% actually write a small check. So it's. I don't know if it's like an equivalent to selling a particular product but I do feel like having product market fit or like find like spending time to build your own product to a degree that like the market would just ask for it without being very salesy is like very important. And then the other not like super contrarian take or like actually feel like it's kind of contrarian as well is by. It's like from the CR of Nidias like Mark, he was president of Twilio and well a former CRO of Twilio and former president of Cloudflare. And so I think one of the things that he said was like if there's no big tech company competing with you, then you're not building something super substantial because it kind of like validates your market, it validates your team and then just like kind of like it's also like a market that kind of like Just like worth fighting for. I feel like those are like the contrarian take in terms of like the macro on the macro level, basically. So those are like things I felt like were really interesting to learn from them.

Yeah, it's kind of a non contrarian, contrarian perspective.

Yeah, non contrarian, Contrarian perspective.

Yeah, I like it.

It's very meta.

Were there any like, companies or demos that kind of made you rethink, like, how AI will, you know, evolve over time or like, gave you a new perspective on like. Oh, that's. I hadn't thought of it from that perspective before.

I like, I feel like, not necessarily contrarian, but I would say, like, I can summarize as like a really big trend. I think we could start from like the consumer world famous voice AI. AI is that called Note LLM? Like, I'm pretty sure I have it on my phone, but like, basically from Google Notebook LLM. Yeah, yeah, exactly. So I feel like that's like a very interesting phenomena as well. So. So basically people wanted to like, you know, finish their email when they're driving or like trying to learn something by listening to it, but like, just like dumping it into somewhere that can like create some sort of like, voice file. So I feel like voice AI is like, definitely a huge trend as well. So I think that. So the. To summarize this, I feel like people started trying to spend less and less time in front of their computer, but they're trying to use AI to help them to get there.

Well, that's the big guy, right? Is that AI can take care of a lot of things so we can have more experiences. Right.

I just want the AI to take care of the laundry. Basically. At this point, I just feel like the only thing AI cannot do is.

Like anything physical for Tesla's robots. Right?

Yeah.

We'll take the Risbot and tell it to. To do your. Your laundry.

Oh my God. What is it called? The Resbot?

The Rizz bot? Have you seen the Resbot on TikTok?

No. What is it?

We'll put, we can put a clip in here in case you're watching on YouTube. But it's one of those humanoid, like, robots and they've given it a voiceover and they've programmed it to troll people in Austin, Texas. So walks around the street. It has this really annoying laugh and it's got millions, millions of views. There was actually one at the, at the Inbound Summit by HubSpot that you attended with us.

Oh, my God, that's intense. Okay, that's very Interesting. Why Texas? What happened to Texas?

Well, I just think they're based out of Austin. I think the robot cost, like, 40,000 bucks. Someone had some cash to burn. But it. It laughs at people. It flips people off. It'll fall down and stand up. It'll talk to people. It'll quote the office.

Oh, cool.

I think. I love that it's got. It's got digits. So I think. I think it's getting close.

Okay, that's exciting.

But going back to the comment that you were talking about automating workflows with voice, the. One of the AI leaders in Utah that went hiking with this morning, he. He constantly. He's telling me about his workflow where he's built a home server, and he'll drive to work and he'll tell it to code something. And he drives home, he tells it to code something, and before he goes to bed, he tells it to code something. And I haven't actually measured his output, but of people I've talked to, I think he's legitimately making this happen. And his focus right now is we're so focused on perfect. Don't expect the AI to be perfect. Just get it to be close. And often close is better than spending two weeks coding it by hand.

Hmm. Very interesting. Actually, speaking of Texas, or speaking of, like, the getting it too close, I kind of believe in that. But on the other hand, like, when I was, like. So I was, like, actually moderating the i5VC summit and the R panel, there was, like, the former Cof BigCommerce. He has, like, a really interesting perspective because he used to work at PayPal as well. So he was saying that, like, AI to disrupt financial industry could be very hard because it needs, like, a lot of accuracy, but for healthcare or like, some other industry, it's, like, a lot easier. I personally kind of, like, see my. I feel like that's, like, a really good perspective because, like, not everything could be fully done by AI. I feel like if they fail at, like, folding my laundry, that's fine, but if they, like, mistakenly transact my money somewhere or, like, you know, diagnose me with, like, some disease I don't have, I will be very sad. So I think, like, there's definitely industry that needs, like, the accuracy, but there's, like, some other industry that can just be okay with the accuracy, but, like, more focus on efficiency.

I wonder if that goes back to the comment I made earlier, which is like, we're not quite there. We're just like, anybody can build an app because they don't. The AI is good but it's rarely that it gets it to like a hundred percent. And if you're like John's friend, right where you're a senior dev, you're like this is great. It's like saved me like tons of hours and I can totally fill in the last like 20% gap. But if you're a junior dev or you're non technical like that last Even if it's 5% means your app doesn't work. Right. It's almost like self driving cars. It's like well how come we don't have self driving cars everywhere? And it's because like getting to a hundred percent, which is really what you almost like you have to have 100%. Right. And you have all these corner cases is just like incredibly difficult to achieve.

So for sure today we had an AI Utah demo day. I probably should call it more show and tell. We had a Stanford PhD present what they're doing with Smarter DX. And you know when AI first came out I was thinking you know there's going to be an AI doctor who does everything. But they've just narrowed it down. They only look at the charts that doctors provide and they believe they can provide magnitudes order greater on accuracy. But I think that's we're just seeing it just being very, very narrowly applied.

Yeah like sector specific AI companies. I feel like that's very a huge trend. Like I think almost like every industry has like an AI agent these days from like you know, dentist to like lawyer to like anything. Like almost every sub domain expert has like an AI agent for it. When I, when we're evaluating the companies for our list, especially the customer service segment, it's like a very hard competition as we will see. Like so many people are competing in this space. Same as like developer tools. I think those are very popular categories as well.

I've thought about those spaces for years. After my last startup, before codebase and after TinyTorch, I'm like how do I do AI based customer support?

Oh yeah, what's your last startup?

Tiny Torch. We were a hootsuite competitor and Facebook turned off the API and killed us. But at that startup I answered about 20,000 customer support tickets. And so I had this whole vision, a workflow of like how it would actually when you submit a ticket it would pay attention to like what page was was there. It pay attention to the contacts, it pay attention to frequency.

What is that?

There's a lot of things that Zendesk just wasn't doing.

I'm just curious about like in that domain, like do you guys expect there will be like one winner that takes all, or do you guys expect something like, you know, a lot of people would just be building the space and then there will be like a niche cost versus agent for everything.

I think if you look at the main markets, what you're seeing is there's not one AI agent that takes all. And every time they're making an update, each one is just marginally better. So I think most AI is beneficial when it's specifically applied to a specific problem set. And the more specific it is and the more you can train it, it'll significantly outperform like ChatGPT.

So like then that like anyone that's building a company solving for that is ultimately going to fail because the companies themselves will just build their own tool and that will be 10 times better because it's based on their own data set.

I heard it's like, like I think like there is, I think that it's just like a lot, a lot of work to build their, your own tool. Like I've chatted with like the customer success startups or like customer success AI startups. I heard people were like, a lot of people are just uploading their own data to these like kind of tools. So not necessarily like they're building their own tools, but they're just using their own data.

Yeah, but, but that's my question though is because to the earlier part of the conversation it's like as it gets easier and easier and easier, at some point do I not just build my own tool? Because it is literally like, you know, over the weekend I build the tool, I bring all my data back over and now I'm saving who knows how much. Right. That I would have been paying to another company who's just paying ChatGPT or Claude or you know, and marking it, marketing it up. Right.

So I definitely feel like for smaller company that's definitely possible. I know like one of our dinner guests, he is building basically chatbot for the customer service or like chatbot for like AI agents. So basically they're, I think his biggest clients are all the people who have tried to build chatbots on their own in their own company because there's just like a lot of like different moat for that or like there's a lot of like technical difficulty to like got it like right. So I'm not like super technical. So I'm not exactly sure if that's the case. But I feel like for like I see like in Silicon Valley each like I feel like the non super core function kind of app people are still going to buy for it like or like buy it then versus like if it's like a core function to your business, they're going to build it. That's my prediction.

Yeah. Interesting.

Yeah, I mean see a specific application that they train on your data is what's probably going to happen.

Yeah, well maybe too it's one of those things. So I think like the, the big value out of AI is the ability to do personalization at scale. If you boil it all down like that is in my opinion like the really big unlock and it may just be that these companies that are building it for other companies, right, whatever that product is, if they can build in the ability to make it highly personalized on every single customer based on their data, but also based on like their tech stack and how they operate and the types, you know, the, the distribution of their employees or their go to market strategy, like all these little aspects that, that impact your product in some way, shape or form. Then it maybe gets to the point where it's like yeah, like we're, we're just not going to build something that's like that customized for us and this is already that customized. Let's just keep going with what we've got. Right. But I think to the extent it's like enterprise SaaS today, I think that the real difference between it and AI is that even if you build like an incredible product, you're only going to be like, I don't know what like 70, 80% complete for any given customer that you might have. Right. Like it's only going to be like 70, 80% of like relative to perfection. And so there's always room for like a slightly better product that meets a few more needs of that customer. And my hope is that like that's where AI really has an opportunity to step in and take it from 70 to 80% all the way up to like 100% or close to 100% value add. I don't know, we'll see. It's a fast evolving market.

Beyond chat. Grace, what do you see as the next interaction model, like the next major shift in ux?

I think it's becoming chat as like you know every from like a fashion app to like these any other like from like booking your dentist appointment. I feel like people are implementing chat into it. So like one example would be like there's a company called Ulta, they do essentially fashion recommendations. So like for example what you're gonna wear next week. And essentially I feel like it's kind of like a chatgpt for fashion. So like you kind of like tell them, hey, I'm going to a ball or like I'm going to like a party, I'm going to like, you know, plan my next week's like work outfit. So essentially the primary function is to like chat with the bot to tell them your request and then they will like recommend you something. So I feel like the next interface or like the next UX is like primarily going to be chat. There is also other companies that are building in like a similar category. So for example, I'm an investor of a company called Generative Alpha. So essentially they're chatgpt for finance Bros. So like if you live in, if you're like working at a hedge fund or if you're working at any private equity firm, so there is always like the analysts that are essentially like researching a specific subject and doing all the research, pulling data from financial reports and that company is building something to replace the human agent to do the report for you or like to do the basically the heart research for you. So I feel like a lot of the newer interface is a chat equivalent function.

Do you worry at all or how do you think about or what have you heard in terms of. There's all these tools that are like, yeah, we're going to get rid of like the junior people. And the thing I wonder is like, well then how do we ever get senior people?

Yeah, so there's actually people training for that. Like, so I actually heard about. So like I was, I had the managing director of Merkor on our podcast like on Monday. So I, I think I asked something similar, but not just like his answer, but I guess like from a lot of people's answer was like there will be companies that are doing the training. And I feel like it's kind of backwards because technically like these will be the junior people on your team to do it. But I do think like it is one of the biggest fear for people to become like dumber as like they're very dumber and lazier because they're super relied on, you know, AI to code for them or like AI to do a certain like marking script. So like there won't be any kind of like junior people right now, so there won't be any senior people later.

So people are worried about that. But you think it gets solved with training?

I don't know if personally I agree training is the best way. I still feel like people need some sort of like monitoring to AI. But I don't, I don't necessarily feel like you should still. People train, you should still train people for the same skill. Unless I feel like you're like a doctor that's actually performing surgery on people, I think you should try immediately. But I think like the rest of the like, like business related things, I think you just need a different set of skills to like kind of like solve those problems.

Yeah. So it's interesting. We. So at my fund, University Girls fund, we hire a ton of interns. That's a big part of like what we do is we teach students how to do venture investing. And part of that application process, we have them do a case study. So we give them like three companies in a week, we don't give them any real information and we say like, go evaluate this company as that word investment opportunity and come back to us. And I thought like this last semester's cohort would have like the best submissions we've ever seen because like they've got like these incredible tools at their fingertips that like are fully fleshed out, fully adopted, like, you know, in a way that just hasn't been possible before. And honestly, like, I think the applications were the worst we've seen in like 16 years of doing this. And everything like fell on into two buckets. Either one, they didn't know how to use the AI tools and so the end result was like not very good. And then on the other end of the spectrum was they relied too heavily on the tool. Right. They just dumped a prompt into deep research and then dropped it out into a PowerPoint presentation and submitted it. And then when we asked them, we were like, you know, even basic questions like, you know, they would put like, oh, the pre money valuation is X. We'd be like, what is pre money? And they couldn't answer it. Right. And so I don't know, it just kind of made me like really worried from a perspective of like, you need to be using AI to get huge productivity gains because if you're not, you're like behind. Right? So like that's a given. But today, like either people don't know how to use it effectively or they're like overusing it and they don't know how to think critically and learn from it. And I think we need to have, I don't know, some sort of like fundamental shift in how we do education and learning. And I don't know what that is. I'm curious, do you have any thoughts or ideas of how we do that, how we solve for that.

I just like, personally I feel like I'm learning by doing type of person. So like I just always like have the bias towards action. So I feel like they have to learn by just start doing tasks. So this is actually like I have also invested in a company called LD Learning. So like basically what they do is like they use. So right now there's like a lot of like engineers that are, they are very high quality raw talents in Africa. However they are not necessarily super trained for the newest like tools and AI knowledge. And so this company created like a learning curriculum by like basically split the big projects from clients to get them to Africa. So essentially the people who are learning they're also earning from the company. So like I know it's like it's con, it's a concept that's similar to necessarily like you just like kind of like hire an engineering crew but they're not really hiring anyone but they're just like paying people to learn. So I guess it's also some sort of like framing or like marketing effects that like people are both learning and training themselves and then like working on these real life projects that buy the company's clients. So I feel like it's like a really good way to approach it as like people need to start working on actual tasks. So what does that meant for your program? Maybe it's like you ask people to source like 10 different companies and like you know like creating memos like to instead of like just use chatgpt but also like kind of like presentations on like how they could leverage AI to kind of like get what kind of information, how do they do diligence beyond the ChatGPT and did they call someone, did they kind of reach out to the founder? Like you can evaluate the ability or you can kind of like train them in a different direction. More focusing on the human side of things or more focusing on the result on outcome set of things instead of you know, I guess like be doing a quiz on like what's like pre money but I guess like more towards like the outcome driven training basically.

Well yeah, it's less for us about like quizzing them on a specific term. It's just like hey, if you're going to put that on your presentation you probably ought to know what it means, right?

Yeah, yeah.

And part of it is like like we, we're not looking for people to have the right answer. We want to understand how people think. Right. And, and how far did they go? Right. To your point, did they call somebody, right? Like, we have applicants in the past where, like, they called 150 retail locations for a product company. Right. Like, yeah, we're going to take that person. I don't care what, like, what the rest of their analysis looked like. Like, that's some real hustle. Right. So, but it's almost like we're not getting any of that now. Like, and, and that's the thing that, that concerns me. But.

So, like, people are a lot more, like, lazier and like, not doing anything. Yeah, I feel like that is very concerning to a degree.

I think we're at the stage, like, with your Goldman Sachs, you could create a custom LLM that trains potential hires, and that's how you get through the junior talent problem. You can actually run them on problem sets and see who performs the best. Yeah, like, or you could do the same thing, Peter.

I just feel like they probably don't need to, like, like, they don't need training for deep research. That's like, at least, like, not in my. I feel like that's like, a good thing. As, like, if they're like, already using AI tools, then, like, they just, like, do absolutely nothing. I fear if a person just literally came in with their own, like, I like that they will, like, think very differently than the rest of the society. But I don't think it's like, necessarily the worst thing ever if they know how to use deep research.

We were talking about the, the future of the younger generation. How do you get jobs in AI? And I think there's an empty space that's opening up. And instead of it being universities to train people AI, you can train them on a custom data set for your type of business. With Carlos Kemeny this morning. He's got two PhDs, and he was telling me how he can create math problems and whatnot, how he's taken an AI, he's taken all the PDFs, he's like, we're going to make these math problems fun specifically for my son. And my takeaway from that was, well, why could Goldman Sachs not do the same type of thing? And they could also make it publicly available like Harvard's courses are. If you want a job here, we are going to teach you not how do deep research, but analysis, but how to approach problems. And if you're not good enough, we can train you on it and see how you're going through, you know, our, our custody, our custom course. And they could also select candidates that way.

I think, though, like, there's, there's an element of like, well, if you're Goldman Sachs, how do you know who to take? Right? And, and so they.

What did I tell you?

Yeah, but this, the AI has to go off of something, right? So I, I don't know. Like, it. That that's part of the challenge is like, it's not that we look for perfect answers in our recruiting. We look for people that are passionate and that are smart. Right? Like, demonstrate that to us through this case study. And I 100% agree, Grace. Like, you need to learn how to use deep research. Like, like it's incredibly important. I also think that like my prompts and deep research for things related to like tech and venture are. And maybe I'm biased here, but are a lot better than what my students prompts are for like the same thing. Because I have all of this, like, I've got like 16 years of doing venture investing behind my belt, right? Like, and there's a lot of context and experience that goes into like crafting that prompt just the way I want it and then engaging with the AI to pull out like the really important things that you. I think it's hard to replicate if you're not like getting real experience. You know, you're in the trenches getting real reps.

But I feel like that's the beauty of like, what you can coach them just because, like, if they. Everybody is like, equipped with like the same amount of information and experience, it like will be living like a way boringer university or like universe in general. Like, I feel like it's good that people can explore on their own as well. So like I feel like it's. Yeah, that's just like my point, basically.

Okay, well, as we're wrapping up, let's do a quick lightning round and then give us a shout out about like, you know, how they can subscribe to your podcast, how they can follow you more. So let's do that. So what's one belief in AI that you will look that you think will. That we believe now that will obviously be wrong in one to two years.

Oh God. My belief in AI. I don't believe in like the there. I mean, I believe there will be like a hardware that powers AI, but I don't think it's going to be the current solution that we have, which is like pin or glasses. Basically. That's just my personal belief. I, I think those are great, but I don't, I'm not sure if like I'll see everybody's just wearing like a glasses. That's like, just tells you what's the.

Weather and stuff like, like what Cluly's doing.

I don't want to point names. I don't think Cluly is telling you about the weather. I think it's just general. I also don't think they're doing glasses. Anyway. I'm just saying any kind of hardware currently, like I feel like the current hardware solution is probably not the.

Yeah, so I, I don't know how they're doing. Just in one of their ads they're showing someone being on a date and like prompting that someone through that date. Oh, you like this? What's your background?

The founder was on our podcast as well. So the founders. But I feel like they're genius at like marketing but I don't think the product was a glasses.

So yeah, I don't think.

But okay, so you don't think like the current hardware solutions that are being promoted are what's going to be the hardware solution?

Yeah, just my personal opinion. I could be completely wrong. Like maybe like two years later everybody's just born in the glasses.

All right, if you, if you have 10 million to invest this quarter, what missed price areas would you back?

That's a great question. I think like this may be a recency bias because I was just on a call with someone who was doing like 3D printing with AI. I feel like that's very interesting. Just maybe like any hardware related things. I think I would back because I think in the end of the day I like to implement AI into our day to day life. Uh, you kind of have to have an entry point from a physical perspective.

Yeah, I think, yeah, I love that to like get to that like next level. It's like your laundry, right? Yeah man, that would be so nice.

I, I just, I just don't need more like you know, voice recorders, like marketing tool. Like I feel like ChatGPT is perfect marketing tool. I just don't need like another like you know this is how we supposed to turn your podcast into a blog and stuff.

Yeah.

Okay.

No offense to like anyone who's doing that. You know if you're really good please like reach out. But if you're like tax then chatgpt, please like reach out.

But okay, what's your favorite negative signal that secretly predicts success?

Negative signal?

Yeah, negative signal, Negative signal.

I think maybe adhd, like it's like a perfect thing for venture just because like if you're like so obsessed with one area you're gonna miss the rest of the innovation.

I think that's interesting. One of, one of my thoughts last night is have I risen to the level of my adhd? Like, I feel like from my. Throughout my business career, as you're talking just about ADHD and predicting success. Yeah, I feel like I'm really good at a certain point, but getting past that point, I feel like ADHD is a curse.

Well, I think, like, okay, so I think, like, to do, like, a lot of things in one area towards your, like, a specific niche, I think that's very important than, like, you just change where you're going every day. That's a problem.

I think if you're adhd, it's easy to be a product leader to an extent, but a certain point you have to switch over to an operator role. And so that makes it tricky. But I, I was just looking at what, what is seen as a negative signal that predicts success. Peter.

I was just gonna say people at adhd. And like, I have. It is the advantage. I think, like, all innovation comes from the intersection of two places. And if you're just constantly curious about lots of different things, that's where you find these, like, corners of innovation that are really interesting. So I like that. I like that insight.

Love that. Okay, Sweet.

Well, Grace, this was a super fun conversation. Thank you for joining us today.

Thank you so much for having me.

So tell. Tell our audience how to. To find more Grace. Goodness.

I am on LinkedIn. I yelp on LinkedIn every day. I have a daily show called Venture with Grace. You can subscribe on Apple Podcasts, Spotify, YouTube and everywhere else. And then I have a podcast called Smart Venture Podcast, which is my original podcast with a little tech executive. You can find it on also Spotify, Apple and YouTube, like, everywhere.

We'll include the. The links below for this episode. So if you're watching or listening. Want more? Check it out because she's amazing. I subscribe.

Thank you. Thank you. I'll like, put this episode on my website as well.

Awesome.

Well, thanks.

Thank you.

Thanks.

Great. Appreciate it.

Thank you guys.

So.