This Week in Startups - Stanford & Google’s Westworld-like AI experiment, BloombergGPT, SF CRE collapse | E1718

Episode Date: April 11, 2023

Jason breaks down an AI experiment conducted by Google and Stanford researchers, where they created a simulated Sims-like video game with characters driven by AI (2:27). Then, he discusses Bloomberg announcing its own financially-focused LLM and what it means for analysts (21:58). He wraps with two quick hits about San Francisco’s dire commercial real estate situation and the collapse of an instant delivery startup (37:51). (0:00) Jason kicks off the show (2:27) Stanford and Google paper "Generative Agents: Interactive Simulacra of Human Behavior" (10:10) LinkedIn Marketing - Get a $100 LinkedIn ad credit at https://linkedin.com/thisweekinstartups (11:38) Prompting the agents and programming AI (20:28) Pilot - Get 20% off the first 6 months at https://pilot.com/twist (21:58) Bloomberg releases BloombergGPT (32:35) House of Macadamias - Get 20% off at https://houseofmacadamias.com/twist by using code TWIST20 (34:03) The future for financial analysts (37:51) San Francisco's commercial real estate debacle  (47:03) MilkRun shuts down FOLLOW Jason: https://linktr.ee/calacanis Subscribe to our YouTube to watch all full episodes: https://www.youtube.com/channel/UCkkhmBWfS7pILYIk0izkc3A?sub_confirmation=1 FOUNDERS! Subscribe to the Founder University podcast: https://podcasts.apple.com/au/podcast/founder-university/id1648407190

Hey, everybody, we have a ton of news for you today. First up, there's a big story out of Stanford researchers have released a paper in which they basically created Westworld, if you've ever seen the show or the movie, or read the book by Michael Crichton. Inside a Sims-like sandbox video game, they created a bunch of NPCs, had them talk to each other, emergent behavior came out of that, so we're going to talk about what Google and Stanford researchers just released. It's pretty, pretty wild, folks.

and it dovetails with simulation theory and blade runner as well. Then we're going to talk about the commercialization of large data sets. Bloomberg has announced their own chat GPT competitor, essentially, their own LLM called Bloomberg GPD. And we'll talk about what this means for financial analysts and figuring out market sentiment and how these data sets are going to be verticalized and will it get rid of tons of jobs or just augment people, automate people, or deprecate jobs. We're going to talk about that very important issue and that transition from augmentation of human intelligence and performance, automation of human performance, and then deprecation of humans themselves.

Finally, two quick hits on San Francisco commercial real estate. Turns out some people have some interesting ideas about transforming Soma into a new type of hub. I think it's a brilliant idea. It gets Mayor Jason's endorsement. An Australian startup has blown through $75 million in one year and shut down. it's in the 20-minute instant delivery space. And so we'll talk about founders creating things that maybe consumers don't need slash are unwilling to pay for.

It's going to be a great solo-dolo show. Please stick with me and us. This Week in Startups is brought to you by LinkedIn Marketing. To redeem a free $100 LinkedIn ad credit and launch your first campaign, go to LinkedIn.com slash this week in startups pilot. grow your business sustainably and operate more effectively. Pilot provides the most reliable accounting, CFO, and tax services for startups and small businesses.

Head to Pilot.com slash Twist and get 20% off for the first six months. And House of Macadamias is the next big health trend. Get 20% off your first purchase and a free bottle of cold pressed macadamia oil at House of Macadamias.com slash twist by your first purchase. using code twist 20. Hey, everybody, welcome to this week in startups. This AI news is coming fast and furious. It seems like everybody is making tremendous progress every day.

And so we've got to keep up with it here. Today, researchers from Stanford and Google did an interesting experiment with NPCs, non-playable characters in a Sims-like video game environment. Again, non-playable characters is what NPC stands for. If you play video games, you know this because they're the people who, you know, video game might be like the shopkeeper, they sell you armor, something like that. And the term NPC has become an insult online for somebody in social media maybe who, you know, doesn't have really intelligent takes. They're not important in the world. Putting that aside that people have

taken the video game term and applied it to the real world, let's talk about what Google and Stanford released today. It's a paper that details this experiment they conducted. The title of the paper is generative agents. Interactive simulacra of human behavior. Interactive simulacra. Simulacra are defined as, quote, copies that depict things that either had no original or that no longer have an original. Basically, these are like Sims characters if you ever played that game, and they're in a sandbox, and they named the sandbox small vill. Here's a screenshot if you're watching us at YouTube.com such this week in. It's a small 2D village. It sort of looks like an old Pokemon game game. 2D. You can walk around. Here's a supply store. The park houses, a college dorm.

And you know, you've played these like 2D simulations. And inside the sandbox, the researchers created 25 characters. Agents is what we might call them. Just like you might consider a chat GPT and agent. You know, you talk to it in a chat format. So these were trained to talk to each other. And then they were trained to have memories of those conversations and then reflect on them. So the researchers used this architecture that stored and blended and applied all these memories. Okay, so now we're getting into like Blade Runner territory, right? Replicans, memories, are they real, et cetera? In order to generate behavior that was believable, they used a large language model, right? Large language models are what Open AI and other folks

are doing. The researchers in this case connected their architecture to open. OpenAI's GPT 3.5. And amazingly, they weren't even using GPT4 to use us. And GPT4 is just a magnitude better. But this is kind of a simple concept here. The architecture was comprised of three main components according to the paper. Memory stream, a long-term memory module that records in natural language a comprehensive list of the agent's experiences. Okay, their memory stream. Reflection, which synthesizes memories over time, enabling the agent to draw conclusions about itself and others to better guide its behavior. And then, three, planning, which translates those conclusions and the current environment into high-level

action plans and then into detailed behaviors for action and reflection. So this is just like, if you've seen some of those simulations, we're like, hey, write me a movie in the style of Quentin Tarantino where five people are bank robbers, they're collaborating on robbing a bank and things go wrong and give me the dialogue. And then at the end, three of them die. You know, and you say, do it in this voice. And it's kind of similar to that, right? Except they've organized it into memory, reflection, and planning. Really interesting stuff. So these reflections and plans are then fed back into the memory stream to influence the

future behavior. Just like when you're using chat GPT, you have that thread and it organizes your conversations with the language model so you can pick up where you left off. And the conversation isn't starting from zero every time you do a prompt or, you know, what we previously would call a search on Google. These prompts now have a history to them. and the history impacts what happens in the future. So here's a graphic they shared of how the memory architecture works.

And so what we're seeing here is there's a memory stream that you can retrieve memories from. You perceive the world. Then it goes into your memory stream. Then you retrieve memories and then you can go act. But retrieve memories could build a plan that then goes back into the memory stream. You reflect on it. And I guess the idea is that this would create more high-level actions or more interesting behaviors in the individual. So what people are starting to realize is with chat GPT4, with chat GPT3.5,

with Dolly, and with creating simulations like this is, well, maybe what we do in terms of consciousness and pulling memories from our brain is not all that sophisticated. It is incredible that we can, you know, I could say hickory, dickery, doc, and you do the next line, right? The mouse ran up the clock. Or I could say, you know, a bird in the hand, and then you're ever, everybody's mind right now, hundreds of thousands of people hear this will say, you know, it's worth two in the bush. It may not actually be that special how our brains work. And so, this is what we're starting to realize is we have this corpus of data known as the internet, known as every word ever written by humans is now online, not in books in the library. So when you

have it all stored and then you build these language models, we can start to predict words, We know what tone is. And so now, all of this is coming together. And that is, I guess, you know, as impressive as this is, what we might actually be learning from AI and these language models is how unimpressive we are as humans. And what we thought was this incredible vocabulary we all have and this ability to communicate. It might not actually be all that special. It might not be. And maybe when we hear a Bob Dylan song or a poet.

or a great writer were like, wow, that is the pinnacle. But it seems to me like these language models are going to be able to deconstruct how that happens and be just as impressive as us. So after they created the memory architecture here, they gave the 25 agents identities. And then they gave them a one paragraph identity description in natural language. This included things like the agent's job, their family, relationships, and other seed memories. This is serious Blade Runner stuff where they incepted memories in these four-year lifespan replicants. For one agent named John Lee, they said John Lee is a pharmacy shopkeeper at the Willow Market and pharmacy who loves to help people. He is always looking for ways to make the process of getting medication

easier for customers. John Lynn is living with his wife. May Lynn, who is a college professor and son, Eddie Lynn, who is a student studying music theory. He loves his family very much. He has known the old couple next door, Sam Moore and Jennifer Moore for a few years. John Lynn thinks Sam Moore is kind and nice man. John Lynn and Tom Marino are friends and like to discuss local politics together. So after creating the sandbox and 25 agents, then giving these agents identities and memory architecture, they let them loose. They basically hit the play button on the game. The agents then started sharing information with each other. In other words, instead of us writing prompts like we do in chat GPT4, they said, well, you know, the pharmacist likes Sam Moore and he thinks is a nice man and Tom Morino and him are friends.

They like to just go out local politics. So they gave them prompts. And so once you get them talking to each other, it would be, you know, like the joke when people put two people on a conference call together. They say, hello, hey, who's calling? And the person who puts the two people on the conference call, you know, a pizzeria and a Chinese food restaurant, you know, and they start talking to each other. And it just creates confusion, but a dialogue emerges. It's kind of like those kind of prank. So this is like a prank being played on the agents.

They basically got them started. Will they stop or not? Okay, let's talk about marketing to senior level. executives, not the rank and file, the senior executives, the people who make the purchasing decisions, the people who okay the purchasing decisions. Well, when you're selling something, business to business, B2B, you need to market to the decision makers. The rank and file, they're probably using your product. They've got the free version, but you need to get those decision makers on board, and it's hard to find them. You're not going to find them on other social

networks. And if you do find them on other social networks where people are dancing or arguing about politics, it's the wrong place to meet them. You want to meet them when they're thinking about business. Business equals LinkedIn, LinkedIn equals business, and LinkedIn has 180 million senior level executives, and 10 million of them are C-level. LinkedIn ads is specifically built to reach those folks, and they have a platform that nobody else has. You can reach those eyeballs when they're thinking about business. You can reach them by title, reach them by company sides, geography, et cetera. And because people on LinkedIn are doing business, they're ready to receive your message. LinkedIn equals business, business equals LinkedIn. You know that. Audience is

exposed to brand messages on LinkedIn are six times more likely to convert above average. So here's your call to action, make B2B marketing everything it can be. And get a $100 credit on your next campaign, a hundy waiting for you at LinkedIn.com slash this week in startups, no spaces, no dashes, LinkedIn.com slash this week in startups. Type out those letters, you get a hundy. Well worth it. You get like 10 bucks for every character you type out. Terms and conditions apply because they've given you a hundy.

So the people who constructed this would also give some prompts to get this whole craziness going. For example, one agent named Isabella was prompted. This is by the humans who created the test to throw a Valentine's Day party. So quote from the paper, starting with only a single user specified notion that one agent wants to throw a Valentine party, the agents autonomously spread invitations to the party over the next two days. The agents made new friends, asked each other on dates to the party and coordinated when to show up at the right time. Isabel invited 12 other agents to the party. And then, seven the agents did not show up for the party. Ooh, drama.

Three of them had other plans and the other four didn't show up for no reason at all. Quote from the paper, Rijiv, a painter explained that he was too busy. He said, I'm focusing on my upcoming show and I don't really have time to make plans for Valentine's Day. Yeah, he's a grinder. You don't have time for parties. I like it. Just like human behavior. So here's a graph.

The diffusion path for Isabel's Valentine's Day party. You know, she says, I'm planning a Valentine's Day party. party at Rob's Cafe on February 14th from 5 p.m. to 7 p.m. And she, you know, asks everybody to come. And then they all have conversations with each other. It's a fascinating idea here. It really does open up this Pandora's box of what happens when a bunch of agents, a bunch of verticalized AI start talking to each other. They're given plans. And this kind of dovetails with simulation theory. Some people believe we're living in a simulation and that we all are these characters. We're Isabella.

And that some being out in the wider universe created what we perceive as the universe, the planet Earth, and we're all in our own simulation. This is simulation theory. And so as you watch this, as you see video games emerging, Grand Theft Auto, Sims, GPP4,

keep in mind, and you've seen the movie The Matrix, I wonder if we ourselves are characters in a simulation. And these agents also form new relationships and remember them without prompting. So, Sam does not know Latoya Williams at the start, but while taking a walk in Johnson Park in the simulation, Sam, the character, the agent, ran into Latoya, and they introduced themselves. And they mentioned she's working on a photography contest. In a later interaction,

Sam indicates a memory of their first interaction as we get real. And he said to Latoya, the agent, hey, how's your project going? And Latoya replies, hey, Sam, it's going well. This was not prompted by humans. So once they got this, this thing going, they said, you know, when you see somebody, say hi to them and, you know, it just started reflecting on what they did with each other. So they started coordinating with each other. This gets really weird. They gave Isabel and Maria two pieces of information. Isabella, you will throw a party. Maria, you have a crush on clause without any further instruction. Isabella invites people to the party, decorates the venue and asked Maria for help, and Maria invited clause to the party.

Okay. An interesting wrinkle in all of this. The researchers could confirm that these interactions were not lucky hallucinations, like when they're just mistakes by the language model, because they could go back and check the memory logs for each agent, so they kind of knew how the history had emerged here. The agent's memory streams that would include logs of them interacting, synthesizing, and remembering new bits of information that they had generated. So is this artificial general intelligence? No, obviously not. This is a bunch of vertical AI talking to each other, and then us starting to simulate what our interactions with each other are as humans, and then applying it to language models talking to each other.

So how does this build into the real world? Well, this is similar to what Tesla is doing with their autopilot. They take all the real world data of actual miles driven. They drive all these miles together, and then when they're driving all these miles together, they find, I don't know, a particularly hard intersection. And then they say, okay, here's every piece of information. here's, you know, 10,000 cars that drove through this intersection that, you know, is painted improperly, or maybe there's been like three accidents there that have been recorded by the city.

Let's run a million cars through this intersection. And let's see if a million cars do this and we give them some parameters, different speeds, different levels of accuracy in the lane they're in. Maybe, you know, they drift up to 20 percent. They change their speed up to 50 percent. Maybe a bicycle goes to the intersection random. those simulations that Tesla is doing with their models is what makes their self-driving so good because they don't have, this is one of the interesting things that's happening, according to

the reports out of Tesla. They don't have a lot of accidents. It's kind of hard to crash a car with a bunch of cameras on it that stops pretty conservatively, like the cruise cars as well that are self-driving in San Francisco. These things are very conservative, so they don't get in a lot of accidents. If they do get an accident, it tends to be somebody hitting them, like a bicycle runs into them or car rear ends them. It's not like they're blowing stop signs or red lights.

They're designed to be super conservative. So how do you simulate accidents? You do it in a sim, which is what they're doing here. So what does it mean? It means video games are going to get a lot more interesting. You could create all kinds of unique video games, random video games, where the characters are just emergent behaviors. And you start to think, well, this is like Westworld.

If you've seen the recent Westworld TV show, you know, they play certain scenarios over and over again, people come for a vacation and it's a real world sim. So what if you took this and you combined it with the animatronics, the early Walt Disney animatronics, and it's a small world? I know this sounds dystopian and nightmarish, but this is what Michael Crichton envisioned for Westworld, which originally was a book and a movie that then became the TV series, which is people would go and interact in a world in real life. Humans made out of, you know, biological materials with robots. So you take the Boston Dynamics robots you see flipping around, it's completely conceivable in our lifetime that you could have a Westworld-like simulation

where we go in and we play paintball or play a war simulation with those robots. They're designed to not kill us, of course, but we could fight, you know, on the beaches of Normandy in a simulation with Boston Dynamics robots that are developing. Or it could be something nonviolent. It could just be, you could go to a simulation of a party. And of course, if you can do that properly, well, then maybe when we're old, there'll be a robot in our house that gets to know us

and in our dying days, instead of having a nurse or, you know, an aide come to our house, you could have this robot, again, like a Boston Dynamics robot or Tesla's making a general interest, a general robot. They could be learning about us over time.

They could be given the programming, be compassionate, have an interest in people, ask provocative questions. They might read, they might listen to this podcast. They might read my biography someday or read every email I've ever written and then just start interacting with me with that as the backstory. Maybe they've read all of your social media. They've looked at all your Instagrams. And this robot who's living with you as you are getting towards the end of life is like, hey, remember when you went skiing in Japan in 2023? I was, you know, watching some videos of it.

Would you like to, you know, put on a VR headset and experience it again? Because you seem like you were really happy. It's just wild to think of what could happen here. This could be simulations of proteins in your body, right? Interacting with each other. We're moving into a space where I think we're going to close the loop in a very short period of time, a couple of years, you know, single digit years on kind of how consciousness works and how we and our intelligence works. And so if you've heard the Wolfram Alpha episode of This Week in Startups from a week or two ago, he kind of talked about this. And so brave new world, it's moving faster than I think any of us thought because this technology, specifically what open AIs them, which is one of the greatest,

I would say, top 20 launches in the history of technology. Because that exists, people who are doing this kind of research, it's almost like they have an Amazon web services. They've got storage, they've got compute power, that they can just run experiments on. And that's really the interesting thing that's happening here is because so many projects are hitting critical mass at the same time, the experiments are starting on third base. You don't have to build a bunch of computer, you don't have to buy a bunch of computers, rack them, and run a natural language model against it. You're going to be able to pick from a hundred different server farms with a thousand different AI models on it and then start putting these together. In fact, this could be

put together with mid-journey and they could create new characters, two of the characters could have a baby in this and then it would generate a new image and build the character automatically. This could get very interesting. All right, everybody, in the early days of a startup, you need to be focused on product and customers, but you can't forget about getting your accounting dialed in. As an early investor, I've seen this so many times, so many accounting horror stories. I have people who are doing cash-based accountants instead of accrual-based accounting. They forget to do their taxes, they're paying for things improperly or they're spending money that's not tax deductible, they're misclassifying expenses. And what happens? You're doing a fundraising and it gets

killed in due diligence. You need to get your accounting right from the beginning. So when you're a founder, you have to balance getting product market fit, building your team, but also not forgetting to set up the proper accounting system. The good news is the perfect solution is waiting for you. Pilot is an amazing service that provides accounting, CFO and tax services for startups. Any problem you're having, they've seen it 100 times, they've seen it a thousand times, they've seen it 10,000 times. Let them handle your books while you focus on your startup. And then when you get big enough, no problem, hire a CFO, bring all your accounting in-house. They understand. They're going to help you as you get to the CFO level. When you get the CFO, yeah, CFO is going to take care of

things. You know what a CFO cost? Did you know what CFOs want to work on? 20 million or more in revenue. Okay, so Pilots got to you. you covered. Twist listeners can get up 20% off for six months at pilot.com slash twist. What a generous offer. Pilot.com slash twist for 20% off your first six months. I think an interesting transition is to think about how this is going to, in fact, business in the short term. Well, Bloomberg is building their own LLM called Bloomberg GPT. Two weeks ago, Bloomberg announced this LLM, large language model, right? Guess the next word, kind of of and Bloomberg, as you know, is one of the largest financial technology companies in the world.

Terminals cost about 25K per year per user. This is how Michael Bloomberg made all of his money, but they also have TV shows like their news network. They do journalism. They write stories. And Bloomberg has over $10 billion in revenue according to different sources. A lot of people use these. Their large language model was trained on the company's own financial. data and data from the web. So it was trained on over 700 billion tokens. GPT3 was trained on

500 billion. They did not disclose what GPT4 was trained on for competitive purposes. But about half of Bloomberg's 700 billion tokens were sourced from Bloomberg's own financial data. So they have a lot of little pieces of information because they have trades, right? They have every piece of data on gold and when they traded at what? It took about 53 days. Bloomberg reported to train Bloomberg GPT, 53 days of computations run on 64 servers, each containing 8 Nvidia A14040 gig GPUs. One of these GPUs, these Nvidia A100s that people talk about,

they cost about 12,000. And so they're very expensive. And that means each of the servers cost about 100K. You multiply 100K by 64 servers. It costs about $6 million in hardware just to get this up and running. And this is why InVin Video stock is going crazy. It's doubled in the last six months because people are realizing, hmm, there is actually a market for these GPUs at scale,

and everybody's going to buy hundreds of them and run data against them. But, you know, it's a de minimis amount of money for a company like Bloomberg, obviously. And so the financial datasets were 54% of what they did. So that's really interesting. Now, you know, chat GPT, if you were to ask it to analyze stocks and trades, where does he get its data from? Like, probably like news articles on the web, maybe not the granular data in data sets that Bloomberg has. And also Bloomberg has proprietary data. So you can be sure Bloomberg is not going to allow Google Bard or Bing or ChatGPT or anybody else to use their data.

If you want that data, you're going to get access to it only by having the $25,000 a year thing. So public company filings, which I'm sure everybody will have access to because they're public companies, SEC, that was 2% of the training data. Press releases from public companies was 1% or so. It's really sort of interesting when you think about it. Public data sets were the other 48% right, so private was 52. And the pile is an open source data set that has been used to train. multiple GPT LLMs in the past. So the pile, which I've never heard of, is a public data set, I guess.

And C4 was 20% of the training data. C4 stands for a colossal clean crawled corpus. It's an open source common dataset that's been used to train LMs, and Wikipedia was 3% of it. So there are public data sets that everybody uses, and I think Wikipedia deserves a lot of credit here because Wikipedia really structured a lot of data that is, I think, by definition allowed to be used by anybody because of their Creative Commons license, as long as you give credit. And I think you have to cite. So I think that's going to be one of these legal issues

that'll be worked out over time. So what's this all going to be able to do? Well, you might be able to ask it, you know, which stocks were, which companies were doing particularly well 10 years ago went through a management change and are doing better now and explain how volume has changed over time and tell me about which company you think is going to go through this pattern before. You could say, hey, these 10 companies, Peloton, Coinbase, whatever, give me 10 technology companies that went public over 10 years ago were worth over $10 billion, but went through more than two management changes and their stock prices down 50% from their peak. Okay, now you get that list.

Okay, put it in a table. Tell me the price earnings ratio. Now tell me which one has done lay. offs, right, public news data, and which one has the largest revenue per employee, which one, and then compare that to the ones that have the largest revenue, the largest earnings per employee, not top line revenue, but bottom line, right? So this is kind of weird queries you give that you might say to an analyst in a brainstorming meeting at your financial company, but now you're just going to type it into a box. And then you start getting back this

information. I've been doing this kind of stuff with chat GPT when I did my tweet the other day about what were the greatest product launches of all time. I literally did this. Tell me the greatest product launches of all time. Put it in a table. Tell me the year it was launched. Tell me the company and tell me how many units were sold. Now, it didn't get that perfectly, but it started to talk about, you know, GPS.

It started to talk about the Mac, the iPod, iPad, iPhone, all those kind of obvious things. Sentiment has always been hard for people to understand. Like, are people negative on a company or positive on a company? Why is that? How does it affect stock price? It might be that negative sentiment about a company in the New York Times and let's call it, gosh, I don't want to trigger anybody. But let's say the woke press, the left press, maybe if Mother Jones and the New York Times and MSNBC and certain Twitter handles who are quote unquote woke or, I'm using the modern term mark, not the original term, who are highly liberal, anti-capitalism, pro-socialism or social Democrats, That's even a better way to say it. Hey, what are social Democrats complaining about? That actually might be negatively correlated with stock price. It might be when Bernie Sanders and MSNBC and Mother Jones are complaining about Starbucks, you should buy Starbucks, right? And then, oh, well, what about

financial people complaining about it? People on CNBC, people who are stock traders. If they're complaining about a company, maybe you should be shorting it because they're complaining about the company because it's not actually doing things that would lead to higher earnings. So that's where sentiment analysis could get very interesting. Whose sentiment are we talking about here? Are we talking about people who hate capitalism or who love it? Bloomberg noted that a company cutting 10,000 jobs could be viewed as really negative by the public, obviously. But the market responded positively.

I told you all on this podcast, I made a J-Trade. You can go to J-trading.com to see my trades. I bought Facebook when Zuckerberg announced he was cutting 10,000. people. That Sunday night, I put an order in. It was trading at 91 on Friday. I put an order in on Sunday. You got filled at $94. The stock quickly doubled. But the sentiment of that was terrible. Oh my God, 10,000 people are losing the job. And rightfully so. It sucks when people lose their jobs, but it doesn't mean like, it means the company's probably going to have higher earnings. So this is going to be a complete revolution, I predict. And the AIs are going to tell us things that we may

not want to hear. We may hear things from AI that are very uncomfortable for us because humans have bias. And we might sugarcoat stuff, right, we might not want to, um, admit certain things, uh, about populations, governments, companies, human nature, etc. It might turn out like, you know, in a recession you should be buying cigarette companies, guns, you know, in the stocks of McDonald's or some unhealthy food. And all this data is going to be super quick to get through. So the amount of time it takes for the information advantage that Wall Street might have had now might come down the long tail and be information that we can actually a civilians trade on,

making short bets, you know, who's a really good CEO, who's not. You'd have to do a lot of research into that. You might need to even previously talk to people. Hey, is this person a good CEO or not? The person who took over Peloton from the founder, are they actually a killer? Are they smart or not? You could look at their track record, but, you know, the AI might help you to find more interesting things out about that person. And then you can place a bet on Peloton stock one way or the other.

So this is an incredibly brave new world. I knew that all this would get, I guess the term might be balkanized. These pools of data would be very valuable, and people's access to it is going to be limited. So I could see Bloomberg only being available for Bloomberg GPT. I could see them giving some of their data if it's cited and linked to and has the logo of Bloomberg in it. Maybe they'll sell barred for $100 million a year, access to it, $50 million a year. So this is where I think a great reset is going to occur where people who own content are going to get paid for their content

for the first time. And yeah, I can't wait to play with this Bloomberg thing. If anybody at Bloomberg has access to this and wants to, when people at Bloomberg have access to it or somebody from Bloomberg's listening, can you send me a login, Jason at Kalaghanis.com? Thank you. In related news, I mean, it's amazing what's happening here.

Google's growth stage VC arm, which is called capital G, just let a hundred million dollar investment in Alpha Sense. This round values Alpha Sense at 1.8 billion. It's a corporate data firm similar to Bloomberg, and a portion of that 100 million is going to be used to help integrate LLMs into Alpha Census platform. CEO said that they are working on a feature that quote will automatically summarize financial documents for customers so they can more easily glean key points, basically exactly what Bloomberg GBT is doing. Here we go, folks. There's going to be many different flavors of this. You're going to have to pay for all of them. If you love snacking, like I do, finding the perfect snack is just impossible, right? You want something that's going to remove your cravings. You want something that fits your dietary goals. You want to regulate that blood glucose level for sure. And most importantly, it has to be

something you look forward to eating to, right? It is a snack. It's something, a little bit of a reward maybe. Well, I found the perfect answer of macadamia nuts. Maccadamias have really unique health fits, they are the only nut, rich in omega-7s, which are linked to natural collagen production, reducing inflammation, stabilizing glucose levels, and healthy fat metabolism. And macadamias are the lowest-carb nut. The folks at House of Macadamia are obsessed with making the highest-quality macadamia products possible. Now me, I love the chocolate-dip macadamia nuts. Oh, for me, a much better choice than, let's say, peanut-ev-and-ms, right? Because I love their chocolate. It's a higher-end chocolate and I love macadamias. They're so rich and delicious. And the chocolate coconut

macadamia bar is also amazing. I take that when I go skiing. And they just launched cold press macadamia oil with a buttery flavor and a high smoking point. So it's perfect for cooking or drizzling over your other meals. Here is your call to action. Use the code twist 20 and get 20% off. What a great deal. And for a limited time, they're actually giving a free bottle of that premium cold press extra virgin macadamia oil with any purchase. So all you have to do is go to house of macadamacadamias.com slash twist and use the code twist 20 bigger question what does this mean from financial analysts right what role are they going to play when automation occurs well i i think they'll be augmented and that's typically what happens here augmented human performance right

like a word processor made what we previously called the typing pool or secretaries uh better at their jobs, they could type faster, they could print documents faster. Then you had document management, you know, did QuickBooks or Excel or Lotus 1, 2, 3, get rid of accountants? No, we had more accountants than ever. People just did more sophisticated analysis faster. So what's probably going to happen here is just people will become much more sophisticated and they'll be augmented. But after augmentation comes automation. And after automation, becomes deprecation. So augmentation,

automation, deprecation. So what are examples of that? Well, you used to have somebody who would actually be typing these documents in or you had phone operators. Over time, that got replaced

with technology, right? And we'll see that with 4 million truck drivers and then there's like another 4 million people who support the truck drivers. Maybe there's 10 million people doing truck driving in this country. Maybe we'll see one truck driver

we'll be able to handle four trucks, you know, driving down the road in a, in a chain, right, like the railroad kind of concept. And it'll be all automated. We'll see just much more augmentation and then eventually automation and then deprecation of those jobs. That takes decades to occur. And we saw it, obviously, in people working on farms, different tools to plow the fields. beast of burden, ox pulling plows, and eventually steam engines and then automated tools to do this and put fertilizer into the ground and plant seeds, all this stuff eventually got highly automated and the number of people it took

to pull wheat or corn became very small compared to what previously happened. Here's a quote from very Van Spina. Bloomberg GPT is on Twitter is going to replace the analysts. Analysts are fundamentally chat-based interfaces

that senior finance folks use to gather, organize, and output data. Finance workflows are already very iterative, and GPT doesn't care about protected Saturdays. There you go. So that's what's happening in a little bit of a nutshell here in terms of how fast AI is moving. and it's going to be amazing.

The technology industry is going to go through a total renaissance over the next decade. Unlike crypto, which was interesting, fascinating, intellectually stimulating, but didn't have much of a use case. But besides money transfer and storage of money were value, store of value, maybe in collecting NFTs, those three cases were very interesting and they're very real, I guess. and some number of people, you know, typically people not in the developed world find those really fascinating, or at least two of those, storage of money, transfer. So it's going to be amazing. But one thing that's not going to be amazing over the next decade, so just to contrast that, AI is actually really real.

People are actually using it to get stuff done to GSD, as we say in the industry, get is done, get stuff, use another word with us, for GSD. thing that's really going to be problematic is commercial real estate. San Francisco is absolutely collapsing. And we're wondering, is there a way to fix this? Well, based on Q1 data, commercial real estate, and this is by Cushman and Wakefield, which is a major CRE firm, the report was tweeted by a friend of this pod, Zach Collius. Employee office attendants continues to track at right around 40% of pre-pendemic levels. In other words, 40% of people are showing up at the office or maybe 100%, maybe 80% of

people are showing up for half the amount of time. That seems pretty low. That seems pretty high to me. I don't think these offices are being utilized 40% of the pre-pendemic levels. I'll be totally honest. But this means the vitality of places like Soma, South of Market, where we've tried to seen some pretty horrible violence and, you know, just economic despair as restaurants, clothes, and cafes and bars and stores and retail and whole foods. Everything just kind of just evaporates

because there's no they are there anymore. Well, overall vacancy, not how utilize these spaces are. Remember, the spaces are being utilized at 40%. I think it's less. I think it might be like 20%. Overall vacancy is now at 25%. This is the highest vacancy rate on record in San Francisco. Other commercial real estate firms have reported 30%. So what does this mean? Well, the other, if the 70% is only being used 40% of the time, what's the actual real occupancy rate? I think it's like 20%, 30% of office space is actually being used. San Francisco has 21 million square feet of vacant office space. It's probably double that amount if I, I think that's actually what's happening.

This is the equivalent of 15 Salesforce towers. If you know the largest building in San Francisco, the Salesforce tower, iconic Mark Benioff Tower. There's 15 of those that are empty. And in fact, Salesforce is getting rid of some of that space. But somebody had a solution. an individual named Bilal Mahmood, who was an entrepreneur. He created a dinner club, and in mid-January, Mahmood's club met to discuss how they can improve downtown San Francisco,

according to their tweet storm, one of the guests, Zach Klein, a co-founder of Vimeo, a very famous entrepreneur, suggested the city should create a new university in downtown San Francisco, or a vast array of college student housing in the downtown core. What a genius idea. The theory is, many of the commercial buildings, they're never going to reach max capacity again, which is true. Current vacancy rates, all time high, like we said. But they could be converted into student housing. That might let them get around a lot of the red tape that there is around this. Schools also get large tax breaks and schools like Berkeley in the NIMBY neighborhood of, you know,

the East Bay here, they have been blocking them from building housing. So if you either moved part of the Berkeley campus to Soma, or you just house students in Soma, south of the market area, and just let them take Bart to Berkeley, hey, this could be really powerful. An analysis done by an architecture firm said they believe the vacant offices could be remodeled into 11,000 living units. Now, this is hard to remodel them. Why is it hard to remodel a building? Well, the plates, how the floor plans are done, tend to not have as much water and sewage. They kind of have, you know, big bathrooms in like two places or three places, four places on the floor plate. They may not have them in what, you know, piping and sewage in 40 places on the plate. And they might have a lot

of interior space, which you're not allowed to use because the way humans are allowed to be housed, you need to have a certain amount of natural light. So maybe there could be some, concept here, well, that maybe students don't need natural light in their rooms where they sleep for eight hours. They could be by the windows, and we'll take the Salesforce tower, maybe by the windows could be a common space where you work, but maybe when you sleep, do you need to have a window in your tiny dorm room? I mean, it would be ideal, but would a student care if they slept eight hours without having a lot of light, if they spent the, you know, most waking hours in the lighted area with the beautiful view of the bay? I think we might be able to change some rules.

there? I don't know. The cost of these conversions could actually be feasible. And I've heard this debate. Some people say it's not possible. The people say it's possible. I lived in New York. We basically found any commercial space we could and we converted it. And we dealt with the issues. Like we built illegal bathrooms, like literally tapped into water supplies. We probably weren't supposed to. I saw people tapping into the water from the sprinkler system, hijacking electricity and running illegal cables. Like, where there is a will, there's a way. If we could have been, If we hacked it in the 90s in New York, the 80s and 90s, we could hack it in San Francisco in a legal formal way. So, if this were to actually happen, UC Berkeley is just the greatest candidate ever.

The UC education system hopes to increase enrollment of Berkeley by over 10,000 by 2030, but Berkeley can only house 23% of its students. And so that's a big rub. and people who live in Berkeley don't want more cars, more housing. You know how this NIMBY stuff works. You own some nice little single family home. You don't want to deal with a bunch of student housing.

You kind of resent the university. You don't want it to grow. I saw this when I lived in L.A. in Brentwood. The Brentwood School, private university, wanted to build a lot more. And then everybody complained about the amount of traffic, pickup, drop off,

all this stuff becomes an issue. So construction has been delayed in at the university. They filed a lawsuit against this $312 million project at People's Park in Berkeley. Construction has been delayed on and off due to protest from activists. And let me tell you, there's a lot of activism in the East Bay. They want to preserve the park as a historical site. And anyway, back in February, state appeals court took the side of the activists and halted the projects. spokesman at Berkeley

ensured the commitment to the project was unwavering. And that development would contain 1,100 beds and 125 for the unhoused, aka homeless folks. What does Mayor Jason

think about all this? Mayorjason.com and real estate. It's a brilliant idea. We need to convert as much of this office space into housing as possible. And if we have to

create some new emergency laws, this is an emergency situation. If one of those laws is there'll be some small, what they call micro-apportments or smaller units, should we be a little flexible here for housing for students? Yeah, I think if it's temporary housing for students and it has a one-year reset, great. We don't want families living with no windows, right? That's really quite dystopian. So yeah, there's a lot of nuance here. Nuance is hard, especially in a place like San Francisco where everybody wants to fight over things. And there's the NIMBY contingent. There's the woke contingent. There's the economic contingent, the capitalism contingent.

Mayor Jason thinks all units equal better. And the more utilized downtown is, and the more vibrant it is, I think the more tax revenue will be, then the more police you could have in policing. and it's not a ghost town. And when it's a ghost town, man, it gets scary and then nobody wants to come. You had, I hate to say safety in numbers, but when downtown was packed and people on bar were packed, it just felt like there was, in fact, safety in numbers.

It was more vibrant. It felt less. It felt like there was a less chance of random violence. And I can tell you now, do not come to San Francisco on vacation, go to Napa, go to the East Bay, go to Lake Tahoe, you know, come to Palo Alto, Stanford, San Jose, you know, the wider Bay Area, amazing, but literally going to San Francisco to stay at a hotel is far too dangerous. I would

literally not allow my family to stay in San Francisco at a hotel. I know that sounds crazy. Maybe I sign a historical, but I don't want my parents or my siblings or whatever to risk it. I don't want them to risk it. There's enough nice places to say north of the city, east of the city, south of the city, it's just too dangerous there. And all these bars are empty, no tax base, everything's shutting down. We have to reverse it. It's an emergency situation, take emergency action, and really redo this. In a quick hit story, it turns out this 20-minute delivery concept is too expensive and unnecessary for most people's use cases. An Australian 20-minute delivery service called Milk Run is shutting down one year after raising $75 million in a series A from Tiger. global. Think about that. $75 million to shut down one year. Crazy. The company used a network of

warehouses to store groceries near customers to ensure users would get their deliveries in 20 minutes or less, very much like GoPuff, which is very profitable, had a $14 billion evaluation at one point. They've also had some challenges. GoPuff is still going. They've been on the program, but Milk Run laid off 20% of its staff in February and then just completely shut down, I think, last or on Sunday night. Another example, here we have it, of maybe technologists and entrepreneurs giving consumers something they don't actually need. And this is just another example of the ZERP, the zero interest rate policy environment we lived in, gave a lot of money to a lot of founders to try to do things better, faster, cheaper. Here, is it better to get your groceries in 20 versus

40 versus 60 versus 90 minutes, I don't actually think it is. If you're putting a dozen eggs in your refrigerator, does it matter if the dozen eggs for the week came in 20, 40 or 60 minutes? It actually doesn't matter. Pack of cigarettes, a bottle of vodka, you're having a party. 20 or 40 minutes, does it make a difference? It's nicer to get in 20, but it's not like a disaster. Listen, I'm not into vodka or cigarettes, but for folks who are, can they wait the extra 20 minutes? I'm sure they can. and so this is an example of entrepreneurs just trying to beat the Amazon prime two-day same-day delivery and trying to compete against DoorDash and Uber Eats

and realizing, yeah, we might have overdone it. People don't need stuff this fast. And more importantly, they're not willing to pay the premium it takes to do that. And the premium is having to build these depots very close to people and then have runners at a large capacity of runners willing to grab your vodka and cigarettes and your cord of milk and your eggs and laundry detergent and blitzkrie running to your house. Totally unnecessary. And I think the 20-minute delivery space unnecessary, probably look back on this as like a ZERP phenomenon, the zero interest rate policy

phenomenon. It's not necessary. People aren't willing to pay for it. In fact, people right now are looking at Uber Eats and DoorDash and they look at that bill and they see a 25 dollar tip. I ordered a guy, I had a bunch of people over and I ordered a bunch of burgers from Shake Shack and a bunch of pizza from eight mile pizza, Troy style pizza. And I give a big tip. I gave 22 percent, I think, and it was like 40 bucks in tips for each one because I ordered $200 worth of each. There's a lot of people coming over. And I was like, well, it's 80 bucks. I'm going to be generous to the driver. But then there was also fees in there. And I'm part of Uber one. But some people are saying, you know what? Maybe I'll just.

go pick it up. So I'll order through the app, but I'll pick it up. And I've actually done that. And I'm not particularly price sensitive, but if I'm on my way home from work or I'm out, sometimes I'll just order in the app and pick it up, maybe one out of ten times. So people don't, sometimes you build something that people don't actually need and are not willing to pay for. And there is a limit to what's necessary. All right. Things for you to think about as you get back to your startups today. How can AI, how can these models, help you make your product more valuable to your customers and how can your team internally be more effective? I would suggest you let everybody on your team expense the $20 chat GPT4 and play

with it, get in the plugins, get into the API and start experimenting. And then think to yourself, think to yourselves like what's more important in the world, 20-minute delivery or these AI innovations. Some crypto nonsense, some coin, NFT, Dow project that never materializes, or what we're seeing in AI. It's pretty clear. Mobile, cloud, software as a service, on-demand, GPS, all of this was an incredible revolution. Crypto, not so much, very narrow use case, but AI, very wide use case. Get to it, everybody. All right, we'll see you tomorrow on this week in startups.