The Exchanges

Q Okay, so Tim, help me out here. You started in the down times. I've started in the last month. What does that, what does that mean for me? I mean, where do you perceive our cycle to be now in the industry? Is this a good time to be starting or a bad time?

A You know, the consensus is that Thing to a new normal. And what I mean by that is, you know, Brexit didn't take the world like we thought it would. Valuations are definitely corrected. So it's not just, um, 10 X GMV on, you know, a sharing economy business is how you get valued anymore. People are much more sensitive to unit economics, LTVs, payback ratios on CAC, all these sorts of, you know, sensible ways to value business. And because of that, I actually think it's pretty good time to invest because you also have several massive platform shifts taking off. Whether it's AR, VR, my favorite AI machine learning, you know, the business plan of the next 10 years will be take machine learning AI, add it to existing industry X, disrupt the whole thing. So, I actually think you're starting at a wonderful time because you've got a lot of comparables to look at, as well as a more sane valuation environment.

AI assessment note: “I actually think you're starting at a wonderful time because you've got a lot”

Q So, so, and with, with regards to that and the fighting chance, I'm intrigued as to what industries you think are most applicable and nascent for this vertical. We've seen a, A wide variety of AI startups focusing on the assistants and AI assistants. So I'm intrigued. Around that, what are your most interesting applications of machine learning and AI, do you think, today?

A So there'll be several broad horizontal ones. These will be business unit applications. You've got your customer service, CRM, ERP, those that already have large existing enterprise software businesses for them right now that you could potentially augment or disrupt. With machine learning, AI, customer service, probably the first one I've seen there. We'll probably then also see marketing, sales, CRM being augmented with AI, and then there's vertical applications. These would be your, you know, for example, in healthcare, being able to read radiology charts more accurately with AI, those sorts of things. The big verticals I do like that have big data pools would be financial services, especially things like insurance, lending, but then you've also got the telecom market, which has Large, large data pools on their subscriber behavior and their data usage, those sorts of things, uh, and then eventually retail.

AI assessment note: “So there'll be several broad horizontal ones. These will be business unit applications.”

Q So, so, and with, with regards to that and the fighting chance, I'm intrigued as to what industries you think are most applicable and nascent for this vertical. We've seen a, A wide variety of AI startups focusing on the assistants and AI assistants. So I'm intrigued. Around that, what are your most interesting applications of machine learning and AI, do you think, today?

A So there'll be several broad horizontal ones. These will be business unit applications. You've got your customer service, CRM, ERP, those that already have large existing enterprise software businesses for them right now that you could potentially augment or disrupt. With machine learning, AI, customer service, probably the first one I've seen there. We'll probably then also see marketing, sales, CRM being augmented with AI, and then there's vertical applications. These would be your, you know, for example, in healthcare, being able to read radiology charts more accurately with AI, those sorts of things. The big verticals I do like that have big data pools would be financial services, especially things like insurance, lending, but then you've also got the telecom market, which has Large, large data pools on their subscriber behavior and their data usage, those sorts of things, uh, and then eventually retail.

AI assessment note: “The big verticals I do like that have big data pools would be financial services”

Q Well, thank you so much, but I'd love to get started today by, by asking my favorite question of all, which is how did you make it into the wonderful world of VC and to where you are today with Mayfield?

A So I have a confession. I fell into venture capital when I was at Stanford business school and I didn't know about the industry before I got there. Quite frankly, during my time as an MBA student, people used to say venture capital is one of the hardest industries to get into, and it's the hot thing to do, and so I basically said, well, that sounds like a challenge. I wonder if I could get into it, and that is possibly the stupidest reason to get into venture, but it is the honest truth, and it turned out to be a good fit because I had, you know, a master's degree in electrical engineering, which is often one of the prerequisites, and worked at large companies like Gateway and General Motors and the technology side and product marketing side, so that was helpful. And then I had some time in entrepreneurship doing some stints as an EIR at Idealab, so I checked the boxes as an associate. Qualification for venture capital firms was recruited into the industry, but my mindset at the time was, let's see if I can get into it. And I remember my first day of work thinking, oh boy, I'm in the industry now. Now what do I do? Um, keep in mind, this was a prior era when A lot of VCs entered as post MBA associates. These days, I think the qualifications and specs are a little bit different.

AI assessment note: “I fell into venture capital when I was at Stanford business school”

Q Absolutely. I mean, what's your perspective on down rounds and bridge rounds? We see more and more bridge rounds. Is, is that a sign that clearly product market fit hasn't been achieved to you, or do you think it's actually a sign that maybe they just need some more runway?

A It could be both, but, um, I'm going to throw it out there. Let's call a spade a spade, and there is no real seed two or seed three. That's basically a bridge, right? Um, that pretty much means you haven't found proper product market fit to extend over into a proper series A, so we will see a big series A crunch, um, series B crunch, because rounds that used to be done on momentum now will need more metrics behind them to really prove those out, and if you don't get those, you're looking at a bridge round, whether you call it A series A two, a series seed two, those sorts of things, right? So the undercapitalized seed or series A will be the new big risk for founders. It is easy to get that first seed round because there's so many seed funds around and it's, it's easy to do a rolling close and do, you know, a safe note and get your seven 50 K in. But you could say that is a recipe for being underfunded because the milestones you can prove on only seven 50 K might not really be enough to get you a true series A later, especially as more funds are looking for more data. For that true Series A round now.

AI assessment note: “pretty much means you haven't found proper product market fit to extend over”

Q Absolutely. I'm super intrigued, though, to ask, uh, Jason Lemkin, a friend of mine, always says that the best investors, Harry, are those that know the benchmarks for the next round. Is that something that you perceive to be true, in particular, about knowing the specifics of what it takes to reach the milestones to move to Series B?

A It is the best way to look at a deal because really your risk is, yes, ultimately does it succeed or not, but the immediate risk, can you raise the follow on round for the company? And so if you've got sort of benchmarks that you're working towards, you at least know where the goalposts are to be able to raise the next round. Probably the riskiest thing is if you don't know what the business model is and you don't know what the metrics will be, then you're really flying kind of blind in the dark and just hoping that traction picks up and that you'll Be a little tracked around. In certain spaces today, like virtual reality, that can work because you've got a lot of interest from Chinese overseas investors, and there's no revenue or model that needs to be proven. It's just very strategic for them, so that can happen. But in well-known spaces with well-defined metrics like SaaS or e-commerce or other things, you probably should know what your metrics are to hit for your strong Series B up-round raise.

AI assessment note: “It is the best way to look at a deal because really your risk is”

Q of short time availability and busyness and travel, I'm intrigued kind of, and this is just me asking for advice here. What is the food schedule like? How do you optimize the food schedule for health benefit? And then how do you manage to fit in exercise and kind of fitness with such a hectic schedule as many of the startup founders and VCs that listen have the similar schedule?

A So, um, You know, it's always a work in progress. I'm always trying to learn, but I think I have maybe the ingredients for a silver bullet on this. The first is, if you can switch your diet to more of a ketogenic or paleo-like diet, so you're eating a lot of lean protein, you know, a lot of high-fiber superfood type of vegetables like spinach or kale and broccoli, those sorts of things. If you can do it, cut out all processed sugar, cut out carbs, you know, that are starchy and low-quality carbs. So namely, No more bread, no more pasta, no more rice, no more, no more, those sorts of things. Increase your healthy fat intake. These are coconut oils, avocado, et cetera. And then if you can throw intermittent fasting into the schedule, you'll see immediate wonderful results. I've been playing with intermittent fasting on a daily schedule where it's basically 16 to 18 hours off, and then only eating within, say, a six to seven hour window. Call it noon to seven p.m., one to seven p.m., something like that. That's not actually that hard. Think about it. Really, that's just cutting out breakfast and snacks. If you then also add on top of that a high intensity, 30 to 40 minute weight training only schedule on, say, three, four days a week, you will pretty much achieve kind of this, the six pack, you know, Abercrombie Fitch model look. I can almost guarantee it.

AI assessment note: “intermittent fasting on a daily schedule where it's basically 16 to 18 hours off”

Q that. I do want to get into the AI machine learning element, as you said there, but I do also have to ask, in terms of the cadence of investing, does it maintain the same throughout? Some investors Very much air on kind of continuation of the same and a strict cadence, and others suggest kind of variations according to cycles. What's your thesis on kind of cadence throughout cycles?

A Well, the traditional wisdom is, of course, buy low, sell high. Um, my best deals were actually in the post 2008 downturn. You know, that's when valuations were cheapest, and companies that survived that downturn ended up thriving quite well. If anything, probably the mistake I made last time was not investing more aggressively in 2009, 2010. So that is something that we're thinking a lot about. The normal pace for a fund like us, you know, mainstream early stage VC firm of three 50 to four hundred million, we're typically doing 10, 12 deals a year, right? So, um, a mainline VC, we're probably seeing anywhere from 400 to 800 deals a year personally and doing, you know, one to three deals of those per year. That will probably be consistent, but if things really slow down with the economy and valuations come back down, you could argue that you'll see People still investing because they've got the funds, they've got the capital deploy, but evaluations and check sizes go down for number of deals. They might be investing more, especially on sort of recap rounds or down rounds on companies that might have gotten ahead of their skis on valuation, but are still fundamentally pretty good businesses.

AI assessment note: “The normal pace for a fund like us... typically doing 10, 12 deals a year”

Q And as a Series A investor, I'm intrigued, you know, we always hear about the Series A crunch. I particularly now as an investor in London see a big Series B crunch. Is that something that you're seeing in particular as we move up the funding cycle?

A Yeah, I think so. And, you know, for, for us, a lot of it boils down to on Series A to B is scalability. And is there enough data that shows the unit economic model is working? Case in point, let's say you are e-commerce company or an on-demand company or something like that. Typically people used to say, look at my GMB run rate. My top line is growing really quick, but these days people are a lot more shrewd. They're going to double click down and say, well, tell me about your payback on contribution margin basis. You know, best of breed is you're paying back on first transaction, or can you make your payback on contribution margin versus CAC in six months? Can you make two X on your CAC within say, 12 to 18 months? You don't have those sorts of things. Are your acquisition channels For, uh, paid users. Is it scalable, or does it start topping out? After this amount of Facebook ads, it's all those sorts of things that people are really digging into now, as opposed to, oh, wow, the GMV is growing really quickly.

AI assessment note: “Yeah, I think so. And, you know, for, for us, a lot of it boils down to”

Q potential concern for me when I view AI startups is the kind of lack of a standardized and traditional business model because it's such a new vertical. I'm intrigued as to your thoughts on the potential around formulating a standard business model that could be used or whether you think this is very much a Individualized perspective. What's your take on kind of building that business model for machine learning?

A So the most obvious entry point is, can you either generate more revenue for an existing line of business or save immense cost on a cost line? Customer service is a great example. Can you either augment the existing customer service workers so they can support 10 people instead of one customer at a time, or could you flat out replace, for example, the Filipino or Indian call center? Right. So that's a direct cost savings there. The other model would be for AI machine learning on the productivity side, the, what I think of as the, you know, AI augmented worker. Can a salesperson handle 20 times more accounts, or could they write their emails 10 times better than they could before? We've got a wonderful company called outreach.io that's playing in this space. Imagine a machine learning system that is seeing the emails of every salesperson, not just in your company, but across all of them. Seeing the open rates, response rates, even traction to close. So then you'd have a system that could say, wait, hey salesperson, before you send that email, let me tweak it a bit for you. I think I can write it even better than you could. And so that's an example of machine learning stepping in to help an existing worker do their job better than they could do currently.

AI assessment note: “the most obvious entry point is, can you either generate more revenue... or save immense cost”

Q Mm-hmm. Absolutely. Uh, and, and so I'm intrigued that we, after kind of desperately wanting to get into the hard industry, as you said, how was it when you finally made it in and how did you stop yourself from going, ah, now what?

A Well, I entered yet the fall of 2001. So my second week of work was nine 11. This was a very, very scary time to enter the industry because after nine 11, basically it was three years of nuclear winter and seeing startups only go out of business. And that sort of shaded my perspective because I only saw how things could go wrong. And in some ways, I think it was a disadvantage because I'd never seen an upcycle. Uh, when your work is composed of playing interim CFO and helping companies shut down or reduced workforce, you just get a sort of pessimistic view. And sometimes that's a disadvantage because you don't see how things can take off like they did when the web two point a wave took off subsequently.

AI assessment note: “This was a very, very scary time to enter the industry because after nine 11”

Q Mm-hmm. Absolutely. Uh, and, and so I'm intrigued that we, after kind of desperately wanting to get into the hard industry, as you said, how was it when you finally made it in and how did you stop yourself from going, ah, now what?

A Well, I entered yet the fall of 2001. So my second week of work was nine 11. This was a very, very scary time to enter the industry because after nine 11, basically it was three years of nuclear winter and seeing startups only go out of business. And that sort of shaded my perspective because I only saw how things could go wrong. And in some ways, I think it was a disadvantage because I'd never seen an upcycle. Uh, when your work is composed of playing interim CFO and helping companies shut down or reduced workforce, you just get a sort of pessimistic view. And sometimes that's a disadvantage because you don't see how things can take off like they did when the web two point a wave took off subsequently.

AI assessment note: “This was a very, very scary time to enter the industry because after nine 11”

Q Well, thank you so much, but I'd love to get started today by, by asking my favorite question of all, which is how did you make it into the wonderful world of VC and to where you are today with Mayfield?

A So I have a confession. I fell into venture capital when I was at Stanford business school and I didn't know about the industry before I got there. Quite frankly, during my time as an MBA student, people used to say venture capital is one of the hardest industries to get into, and it's the hot thing to do, and so I basically said, well, that sounds like a challenge. I wonder if I could get into it, and that is possibly the stupidest reason to get into venture, but it is the honest truth, and it turned out to be a good fit because I had, you know, a master's degree in electrical engineering, which is often one of the prerequisites, and worked at large companies like Gateway and General Motors and the technology side and product marketing side, so that was helpful. And then I had some time in entrepreneurship doing some stints as an EIR at Idealab, so I checked the boxes as an associate. Qualification for venture capital firms was recruited into the industry, but my mindset at the time was, let's see if I can get into it. And I remember my first day of work thinking, oh boy, I'm in the industry now. Now what do I do? Um, keep in mind, this was a prior era when A lot of VCs entered as post MBA associates. These days, I think the qualifications and specs are a little bit different.

AI assessment note: “I fell into venture capital when I was at Stanford business school”

Q Okay, so Tim, help me out here. You started in the down times. I've started in the last month. What does that, what does that mean for me? I mean, where do you perceive our cycle to be now in the industry? Is this a good time to be starting or a bad time?

A You know, the consensus is that Thing to a new normal. And what I mean by that is, you know, Brexit didn't take the world like we thought it would. Valuations are definitely corrected. So it's not just, um, 10 X GMV on, you know, a sharing economy business is how you get valued anymore. People are much more sensitive to unit economics, LTVs, payback ratios on CAC, all these sorts of, you know, sensible ways to value business. And because of that, I actually think it's pretty good time to invest because you also have several massive platform shifts taking off. Whether it's AR, VR, my favorite AI machine learning, you know, the business plan of the next 10 years will be take machine learning AI, add it to existing industry X, disrupt the whole thing. So, I actually think you're starting at a wonderful time because you've got a lot of comparables to look at, as well as a more sane valuation environment.

AI assessment note: “I actually think you're starting at a wonderful time”

Q Absolutely. I'm super intrigued, though, to ask, uh, Jason Lemkin, a friend of mine, always says that the best investors, Harry, are those that know the benchmarks for the next round. Is that something that you perceive to be true, in particular, about knowing the specifics of what it takes to reach the milestones to move to Series B?

A It is the best way to look at a deal because really your risk is, yes, ultimately does it succeed or not, but the immediate risk, can you raise the follow on round for the company? And so if you've got sort of benchmarks that you're working towards, you at least know where the goalposts are to be able to raise the next round. Probably the riskiest thing is if you don't know what the business model is and you don't know what the metrics will be, then you're really flying kind of blind in the dark and just hoping that traction picks up and that you'll Be a little tracked around. In certain spaces today, like virtual reality, that can work because you've got a lot of interest from Chinese overseas investors, and there's no revenue or model that needs to be proven. It's just very strategic for them, so that can happen. But in well-known spaces with well-defined metrics like SaaS or e-commerce or other things, you probably should know what your metrics are to hit for your strong Series B up-round raise.

AI assessment note: “It is the best way to look at a deal because really your risk is”

Q Absolutely. I mean, what's your perspective on down rounds and bridge rounds? We see more and more bridge rounds. Is, is that a sign that clearly product market fit hasn't been achieved to you, or do you think it's actually a sign that maybe they just need some more runway?

A It could be both, but, um, I'm going to throw it out there. Let's call a spade a spade, and there is no real seed two or seed three. That's basically a bridge, right? Um, that pretty much means you haven't found proper product market fit to extend over into a proper series A, so we will see a big series A crunch, um, series B crunch, because rounds that used to be done on momentum now will need more metrics behind them to really prove those out, and if you don't get those, you're looking at a bridge round, whether you call it A series A two, a series seed two, those sorts of things, right? So the undercapitalized seed or series A will be the new big risk for founders. It is easy to get that first seed round because there's so many seed funds around and it's, it's easy to do a rolling close and do, you know, a safe note and get your seven 50 K in. But you could say that is a recipe for being underfunded because the milestones you can prove on only seven 50 K might not really be enough to get you a true series A later, especially as more funds are looking for more data. For that true Series A round now.

AI assessment note: “pretty much means you haven't found proper product market fit to extend over”

Q Yep, absolutely. And then what would solve America's healthcare woes? You are the healthcare genius.

A I'm not a genius. I know two things that would, but they'll never get passed in my lifetime because people would Throw up at the thought, but the first is treating processed sugar like a scheduled narcotic substance, because let's admit, that's really what it is. Would you give someone a crack pipe and say, please self-regulate? No. The second would be, if you could limit people's freedom to eat what they want based on their BMI, you know, if you've got a healthy BMI, I think you can eat whatever the hell you want, but frankly, as your BMI creeps off and gets really, really, you know, obese because you're eating choices, Should you be really allowed to eat what you want? Are your choices that great? You know, again, no one will ever like these things. They'll say, ah, that's blasphemy, but food is the root of almost everything related to healthcare and our healthcare costs these days.

AI assessment note: “the first is treating processed sugar like a scheduled narcotic substance”

Q And as a Series A investor, I'm intrigued, you know, we always hear about the Series A crunch. I particularly now as an investor in London see a big Series B crunch. Is that something that you're seeing in particular as we move up the funding cycle?

A Yeah, I think so. And, you know, for, for us, a lot of it boils down to on Series A to B is scalability. And is there enough data that shows the unit economic model is working? Case in point, let's say you are e-commerce company or an on-demand company or something like that. Typically people used to say, look at my GMB run rate. My top line is growing really quick, but these days people are a lot more shrewd. They're going to double click down and say, well, tell me about your payback on contribution margin basis. You know, best of breed is you're paying back on first transaction, or can you make your payback on contribution margin versus CAC in six months? Can you make two X on your CAC within say, 12 to 18 months? You don't have those sorts of things. Are your acquisition channels For, uh, paid users. Is it scalable, or does it start topping out? After this amount of Facebook ads, it's all those sorts of things that people are really digging into now, as opposed to, oh, wow, the GMV is growing really quickly.

AI assessment note: “Yeah, I think so. And, you know, for, for us, a lot of it boils down”

Q do want to touch on one specific area that you mentioned earlier that we're both super excited by, I think everyone's super excited by, and it is machine learning and AI. So my first question is, with such a plethora now of startups suggesting that they integrate machine learning into pretty much every aspect of their offering, how do you differentiate in the now vast pool of machine learning startups?

A Okay, so here's the thing, because machine learning and AI is the hot new buzzword, Expect every single startup pitch to incorporate that as a bullet point, just like they did back when, you know, UGC or crowdfunding or any trend of the year was hot. You'll see that in every deck, but the truth is a lot of times it's probably, probably overreaching, if not total bullshit. The way you can suss that out is look at the PhD caliber backgrounds of the teams on their machine learning or AI staff. If they don't come from one of the six world-class centers will be at MIT or, or others. It probably isn't true. It's probably just automation dressed up as machine learning. So that's something we look at is what is the bench strength of the AI machine learning team there. If those people aren't being systematically poached at by Google or Facebook or whatnot, they're probably not world-class AI machine learning talent. For true world-class AI machine learning startups, they're going to need name brand AI rock stars. And, you know, I only use the term rock star loosely in the sense that They are being approached almost like NBA draft picks now. Same salary amounts, same recruiting efforts, and so being able to land those people, that is the new test these days. I like to say in this era where we're going to be building for AI first on a lot of companies, proprietary data pools and AI talent…

AI assessment note: “The way you can suss that out is look at the PhD caliber backgrounds”

Q of short time availability and busyness and travel, I'm intrigued kind of, and this is just me asking for advice here. What is the food schedule like? How do you optimize the food schedule for health benefit? And then how do you manage to fit in exercise and kind of fitness with such a hectic schedule as many of the startup founders and VCs that listen have the similar schedule?

A So, um, You know, it's always a work in progress. I'm always trying to learn, but I think I have maybe the ingredients for a silver bullet on this. The first is, if you can switch your diet to more of a ketogenic or paleo-like diet, so you're eating a lot of lean protein, you know, a lot of high-fiber superfood type of vegetables like spinach or kale and broccoli, those sorts of things. If you can do it, cut out all processed sugar, cut out carbs, you know, that are starchy and low-quality carbs. So namely, No more bread, no more pasta, no more rice, no more, no more, those sorts of things. Increase your healthy fat intake. These are coconut oils, avocado, et cetera. And then if you can throw intermittent fasting into the schedule, you'll see immediate wonderful results. I've been playing with intermittent fasting on a daily schedule where it's basically 16 to 18 hours off, and then only eating within, say, a six to seven hour window. Call it noon to seven p.m., one to seven p.m., something like that. That's not actually that hard. Think about it. Really, that's just cutting out breakfast and snacks. If you then also add on top of that a high intensity, 30 to 40 minute weight training only schedule on, say, three, four days a week, you will pretty much achieve kind of this, the six pack, you know, Abercrombie Fitch model look. I can almost guarantee it.

AI assessment note: “intermittent fasting on a daily schedule where it's basically 16 to 18 hours off”

Q do want to touch on one specific area that you mentioned earlier that we're both super excited by, I think everyone's super excited by, and it is machine learning and AI. So my first question is, with such a plethora now of startups suggesting that they integrate machine learning into pretty much every aspect of their offering, how do you differentiate in the now vast pool of machine learning startups?

A Okay, so here's the thing, because machine learning and AI is the hot new buzzword, Expect every single startup pitch to incorporate that as a bullet point, just like they did back when, you know, UGC or crowdfunding or any trend of the year was hot. You'll see that in every deck, but the truth is a lot of times it's probably, probably overreaching, if not total bullshit. The way you can suss that out is look at the PhD caliber backgrounds of the teams on their machine learning or AI staff. If they don't come from one of the six world-class centers will be at MIT or, or others. It probably isn't true. It's probably just automation dressed up as machine learning. So that's something we look at is what is the bench strength of the AI machine learning team there. If those people aren't being systematically poached at by Google or Facebook or whatnot, they're probably not world-class AI machine learning talent. For true world-class AI machine learning startups, they're going to need name brand AI rock stars. And, you know, I only use the term rock star loosely in the sense that They are being approached almost like NBA draft picks now. Same salary amounts, same recruiting efforts, and so being able to land those people, that is the new test these days. I like to say in this era where we're going to be building for AI first on a lot of companies, proprietary data pools and AI talent…

AI assessment note: “look at the PhD caliber backgrounds of the teams on their machine learning”

Q that. I do want to get into the AI machine learning element, as you said there, but I do also have to ask, in terms of the cadence of investing, does it maintain the same throughout? Some investors Very much air on kind of continuation of the same and a strict cadence, and others suggest kind of variations according to cycles. What's your thesis on kind of cadence throughout cycles?

A Well, the traditional wisdom is, of course, buy low, sell high. Um, my best deals were actually in the post 2008 downturn. You know, that's when valuations were cheapest, and companies that survived that downturn ended up thriving quite well. If anything, probably the mistake I made last time was not investing more aggressively in 2009, 2010. So that is something that we're thinking a lot about. The normal pace for a fund like us, you know, mainstream early stage VC firm of three 50 to four hundred million, we're typically doing 10, 12 deals a year, right? So, um, a mainline VC, we're probably seeing anywhere from 400 to 800 deals a year personally and doing, you know, one to three deals of those per year. That will probably be consistent, but if things really slow down with the economy and valuations come back down, you could argue that you'll see People still investing because they've got the funds, they've got the capital deploy, but evaluations and check sizes go down for number of deals. They might be investing more, especially on sort of recap rounds or down rounds on companies that might have gotten ahead of their skis on valuation, but are still fundamentally pretty good businesses.

AI assessment note: “That will probably be consistent, but if things really slow down with the economy”

Q potential concern for me when I view AI startups is the kind of lack of a standardized and traditional business model because it's such a new vertical. I'm intrigued as to your thoughts on the potential around formulating a standard business model that could be used or whether you think this is very much a Individualized perspective. What's your take on kind of building that business model for machine learning?

A So the most obvious entry point is, can you either generate more revenue for an existing line of business or save immense cost on a cost line? Customer service is a great example. Can you either augment the existing customer service workers so they can support 10 people instead of one customer at a time, or could you flat out replace, for example, the Filipino or Indian call center? Right. So that's a direct cost savings there. The other model would be for AI machine learning on the productivity side, the, what I think of as the, you know, AI augmented worker. Can a salesperson handle 20 times more accounts, or could they write their emails 10 times better than they could before? We've got a wonderful company called outreach.io that's playing in this space. Imagine a machine learning system that is seeing the emails of every salesperson, not just in your company, but across all of them. Seeing the open rates, response rates, even traction to close. So then you'd have a system that could say, wait, hey salesperson, before you send that email, let me tweak it a bit for you. I think I can write it even better than you could. And so that's an example of machine learning stepping in to help an existing worker do their job better than they could do currently.

AI assessment note: “So the most obvious entry point is, can you either generate more revenue”