NEA's Tiffany Luck on AI IPOs, personal agents, and the ROI reckoning

Speaker A: I see you.

Speaker B: Fire and Ash is now streaming on Disney. It's the film critics are calling the best avatar yet. A true epic and completely jaw dropping.

Speaker A: This is the only pure thing in this world.

Speaker B: Return to Pandora on Disney.

Speaker A: It will be an adventure for the whole fam.

Speaker B: And watch the Oscar winning phenomenon at home.

Speaker A: This is sick.

Speaker B: Fire and Ash now streaming on Disney. Rated PG 13.

Speaker C: Hello and welcome back to Equity, TechCrunch's flagship podcast about the business of startups. I'm Rebecca Balon and this is the episode where we bring on industry experts to help us explore a trend in the tech world and dive deep. Tiffany Luck, partner at NEA got her start convincing companies that E commerce was the future. Now she's all in on AI, especially when it comes to the possibilities of magic moments in consumer. She's joining us today to talk about the future of personal agents. Her thoughts on this year's AI IPOs and how startups are stepping in to help enterprises track return on AI spend. Tiffany, welcome to the show.

Speaker A: Thank you for having me, Rebecca. How fun.

Speaker C: So you have been in the industry for a while but you've been a partner at NEA since 2023 right?

Speaker A: Yes, I joined just at the beginning of 23.

Speaker C: Wow, okay. And so then you've been here for probably like a, ah, seriously big growth cycle. I know that you're focused mostly on early stage AI, APIs, B2B SaaS. Previously you were a partner at GGB Capital. What were you doing before you came into the world of vc?

Speaker A: Okay, I'll take you on the, on the, on the very brief, uh, long winding journey so far. So fun fact, I actually started out on the, started on the operating side doing a little bit of everything, a lot of customer acquisition, business development, marketing. And the fun fact is actually I was a very early employee at a NEA portfolio company called Lot 18 which was back in the early days when people refer to New York as Silicon Alley. So it was a wine flash sales startup. Um, if you remember, you know back in the flash sale days when Gil Rue la la, uh, and everything was starting. So it was actually like one of the first vertical iterations of that which is interesting and probably will something we'll circle back to at some point given how uh, big vertical is today. So did a lot of, you know, a lot of everything. I was like the 10th employee there. Then after lot 18 spent some time at Amazon actually in the early days of Amazon New York which was also very fun and you know Amazon had already been around for A while at that point. But it was actually in the early days of Amazon trying to figure out all things consumer packaged goods. Um, so how to get people to actually buy household items like your, you know, very regularly recurring items on Amazon. Just when they were getting ready to launch Amazon Fresh and Prime Pantry and all these different things.

Speaker C: Wow.

Speaker A: And so, so it was a little bit of an interesting time. I would actually go work with the CPG manufacturers. So I spent a lot of time in the Midwest and actually talk to them about how E commerce is the future. Because back then this is, you know, this is now 2012 or so, it really was not apparent that people were going to buy. I mean, it was apparent to me, but I think at the time it was not obvious that people were going to buy these things online. And CPG manufacturers were still very much levered to brick and mortar. Right. They lived and breathed brick and mortar. Anyway, so kind of like preaching about the future of E commerce, which was exciting and fun. Obviously that has very much since played out. Right. Obviously Amazon bought Whole Foods and now we buy truly everything, including like a toothbrush overnight online. But that was fun. And then also before Venture, spent a couple of years at Morgan Stanley in the tech investment banking group, um, in Menlo park, mostly spending time on M and A, but I also did work on some IPOs and other fun things.

Speaker C: Wow. So there's just so many things we can tap your expertise for today. I'm actually curious of digging into the consumer bit because we've talked a little bit about how consumer is about to pop off or it's really seeing like a moment. Right. So what are some of the magic moments of 2026 that you've already seen or that you expect to see? I mean, I know we talked a little bit offline about personal agents, uh, becoming more of a thing. Do you really think that's gonna happen this year?

Speaker A: I think so. I think we're almost there. And I actually love the way you phrase magic moments. Actually something I talk about all the time and something I talk to our portfolio companies though, like, how can you create a magic moment, whether it's consumer or enterprise? Um, my favorite magic moment for me, one that, like, is very visceral, was my first time riding in a Waymo. Like, it's so magical. Right. You get in and I mean, it's kind of like cool watching the car drive up. It's exciting to get in for the first time or actually, you know, your hundredth time. Um, you know, you get to like, personalize the music in the back seat. It drives very well. Um, um. Though I will say it does, like, very much drive the speed limit on the freeways. So it takes. Yeah, you're, like, calmer. I was like, go a little bit faster here. But, like, the whole experience, like, is truly magical. And you're like, okay, I'm living in the future. And, you know, I feel like we have started to see some of those moments happen within AI, you know, especially, like, I would say to me, it's been mostly on the, uh, you know, me as a professional experience.

Speaker C: Right. The prosumer, the enterprise moments.

Speaker A: Totally. You know, like when Claude is helping you kind of put together something for work or think through something. Actually, Claude's very good at, like, you know, how to have hard conversations, like, going back and forth on. On things like that. Um, I would say on the consumer side, actually, AI very good at healthcare, you know, very good at, like, getting second opinions on things. True. But, like, just as a gut check, like, hey, is this serious?

Speaker C: I mean, you're gonna Google it anyway, right?

Speaker A: Totally. I would say, like, very acute on that stuff. And I feel like we're getting ready to have that moment on the personal assistant side. So there have been a few companies that have come out lately, um, like Town and Ollie, and I think companies are very specifically targeting personal use cases and assistants. Obviously, you can use Claude as an assistant. You can use ChatGPT, um, in some ways for this and other models as well, but just the actual packaging around. What does it mean, you know, how do you want to interface with your assistant? What does it mean to actually, like, carry through a task end to end, to have memory over time, you know, thinking, like, what does it mean to me? I would love to be able to actually walk to work and say to, you know, to my phone, but, uh, like, to my assistant, whatever that would be. You know, here's what I need to get done today. And it actually just takes care of all of it. I feel like we're not quite there.

Speaker C: Right. That's the magic moment for agents. Like, right now, it feels like when you try to use agents, it's. There's actually just more friction. And you're like, I have to teach this thing. Thing. And what. I'm, um. Like, I'll just. Might as well do it myself.

Speaker A: Right.

Speaker C: It's like dealing with, like, a clueless intern that's just gonna mess it up anyway. And you're like, no, it's fine. Just, you know, I'll do it. I'll handle it.

Speaker A: Yes.

Speaker C: You know, and so you could see that being.

Speaker A: Yeah, no, no, I was just gonna

Speaker C: say what are like you mentioned Ollie and another company I haven't heard of these companies. What do they do? They're like a rat. Yeah. Town.

Speaker A: So, yeah, check them out. They're um, basically like, try to take care of everything for you. So I, I always think about them as like actually like managing the mental load that we all carry around because I actually think that's the hardest. Yeah, that's the hardest part. Right. It's not actually completing one task end to end. Like. Yes. In some cases, tasks specifically are complicated, but it's actually like the logistics and um, you know, kind of like domino effect of certain tasks and the remembering to even do certain tasks at certain times. Managing calendars, managing personal and professional calendars. You know, in many cases, like Manning, managing family calendars. Like there are just so many pieces. And so I think the magical moment, um, will be when an agent, assistant, whatever you want to call it, can actually translate your mental load and take that away in a way that, where you actually like, believe it. Right. There's like high confidence level there so you don't have to double check everything. M. Otherwise the mental load still still exists.

Speaker D: Yeah.

Speaker C: I guess another like follow up question on that magic moment is, you know, when you think about personal agents, one of the main things that hold people back, at least holds me back from wanting to use any of these tools to like help me with my inbox or whatever, is privacy. It's safety, it's security. It's not trusting that, you know, you open it up, there's more areas to hack and hacking is getting better and it's really scary. Now, as we've seen with privacy for the whole of the Internet, there's like a, there's like a peak, there's like a peak of caring about your privacy. And then it's like if the thing is convenient or useful enough, you kind of just stop caring. Right. Like that's like a consumer thing to do. Like, what do you think will need to happen? Will we need to get to seriously safe security or does it just need to be so good that you don't care first?

Speaker A: My guess is kind of a little bit of both. But I want to say, you know, definitely safe security. And I think that can happen. I feel like cynical. I know, honestly, I am m too, a little bit. And the, you know, wonderful thing, but scary thing about AI is in some cases it's like, you know, embedded in your operating system essentially. Right. Or it's like, with computer use, like, it can actually see everything you're doing. And obviously in some cases people are like, cordoning it off. But, you know, the idea of actually having an AI that's watching me log into my bank account, I'm probably not there yet.

Speaker C: Yeah.

Speaker A: You know, also, I, you know, I tend to, like, definitely be, uh, of the paranoid type when it comes to security. However, you know, I feel like there's a lot of really, really smart people trying to figure this out. And so whether that's AI versus AI in terms of, like, hacking and security, there's a lot of great companies on that front. Um, you know, whether it's like AI specific security companies, whether it's, you know, we also now have a lot of different, um, certifications or standards cropping up. I think maybe you and I chat about a company called aiuc, the AI underwriting company, which is super cool company. They're basically trying to do. I think it was like either Moody's for AI or like a private SoC2 that actually talks about standards, um, and safety. So they have a group of, I think 100, more than 100, um, very well known CISOs who are actually getting together on a regular basis to create these standards. And then certain companies will get their agents certified. M. So who knows if it's the certification route, if it's the, uh, you know, 100% pure security route. I mean, probably both of those will end up being super helpful. Um, but yes, I am hopeful that we get to a point where we feel really good about the security. And it intersects at the time when, you know, this AI is just so useful, you can't not use it.

Speaker C: Oh, you know what? I just thought of one. Let's start a business. Okay. It's a personal security agent. I mean, yeah, everyone gets their own whatever. Every agent you have, you get your own little personal security agent. That won't be expensive to run at all. Um, but that will be just constantly watching. It'll. It'll play like the security guard. It's your bouncer. It's your constant bouncer.

Speaker A: No, we could call it Bodyguard.

Speaker C: Yes. Okay, great. Okay, we'll talk about agent guard more. I will be requesting $100 million seat round.

Speaker A: Um, I know exactly who will find you.

Speaker C: Okay. So, you know, when we talk about consumer oftentimes, I mean, there's all of these great, like, rapper companies or companies that, like, are building kind of the. The last 20%, like 80% is done by the foundational companies in the last 20% or so is maybe done by a lot of these more nuanced startups. But you know, we have to take it to the foundation models because they can always come in and eat the lunch of a startup. And so right now I'm thinking about OpenAI. They, at the time of this recording, they had just filed confidentially for ipo. Obviously they're following Anthropic. If you look at the numbers, it's looking like Anthropic's really, you know, the enterprise use case. OpenAI, the consumer use case. Of course, OpenAI has enterprise customers as well, but they seem to be well known as more of the consumer play. How do you see this kind of playing out on the public markets in terms of investor sentiment? What do you think investors will be looking out for the most?

Speaker A: Yeah, well, first of all, I feel like it's such an exciting time and public markets have honestly been waiting for this moment. Um, especially like the trio, right? Everyone's been talking about SpaceX, Anthropic, OpenAI. You'd even see this in the private markets, right, with late stage funding rounds. Um, you know, there's just been infinite demand for these companies, which makes total sense. Complete game changers, right? We've never seen companies grow so quickly, you know, and obviously like creating real, real business value. So I think first and foremost, like I'm excited, I think the market's excited. You know, I think like it's really good for the ecosystem, it's really good for liquidity. It's just, it's good for so many reasons. Um, you know, I think like separately I think all of these companies are very valuable companies. I think that they will all continue to grow. I think that they all have a real place. If we just talk about like OpenAI and Anthropic for a few moments. Yes, one may skew more enterprise and one may skew more consumer. I think actually both have very good, um, you know, kind of like inroads for both, I would say, and like both have real use cases to be used. In both cases, enterprise and consumer. I also personally think the lines will continue to blur. Um, you know, well, we still talk about like enterprise business, consumer business maybe. Um, but I actually think like, you know, the line is definitely getting much, much thinner. One thing that the markets will at least start thinking about is, um, you know, not necessarily profitability, but you know, just compute cost, um, and not just like necessarily cost, but just the sustainability. Like are all of the components in place that will actually fuel this growth. Not the demand, but like you know, literally everything that goes into actually delivering, um, the amount of compute required. Right. For uh, the demand that is out there. Um, and so I think people will be looking at unit economics, I think people will be looking at everything to say, hey, ultimate, like will this be a trillion dollar company that goes to $5 trillion and in what time period or $1 trillion company goes to $10 trillion and in what time period and therefore like what needs to happen, you know, in order for the business unit economics to make sense? Um, so that at some point there's free cash flow and at some point there's metrics that uh, the public markets tend to expect, you know, especially with large companies like Nvidia and things like that today.

Speaker C: Yeah. So in terms of like valuations and what these companies are worth, I'm curious what you think about this. With Anthropic's last round and their last valuation and what their trying to raise, uh, in their ipo, it feels like a lot of their, a lot of their valuation is based on their ability to access compute. Right. Same with OpenAI. And the interesting thing is that Anthropic is getting their compute from a competitor, which is xai, which doesn't seem to be using their own compute that much, but is starting to sell it. What do you think is going to be more valuable? Uh, the models and then access to compute or owning the actual infrastructure.

Speaker D: This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50 page restoration block or finally break down that long article you've had open for weeks. Gemini in Chrome is here for it, ready to make anything online make sense. There's no place like Chrome. Check responses setup required compatibility and availability various 18.

Speaker A: I actually think all the above. And that's why I think like I really do think that AI is kind of a, um, you know, like the, that phrasing, like a rising tide lifts all boats. Correct me if that's not exactly what I mean.

Speaker C: Yeah, basically, yeah.

Speaker A: It's like I'm like the boats, the sea, the ship, whatever. Yeah, um, you know, like there's just so many good things happening and I just feel like there's value really being created at all layers of the stack. Um, I think at the infrastructure layer for sure, actually the model layer for sure. Um, you know, obviously there's some debate around will the models eat everything or what will be left. Like I think there's real value at the application layer as well, you know, at the harness layer. Like, there's just. It is such a major transition. It's such a major trend. It's such a major everything, like a change in everything we're doing and something so new that, like, there's just. I still think we're really at the early days, which sounds kind of crazy when you think about it, because trillions of dollars of value have already been created, but in many ways it really does still feel like, you know, kind of first, second, third innings.

Speaker C: Yeah, you mentioned something earlier. You said that AI is already creating real business value. Right. And so I wanted to talk to you about that a little bit because the big conversation of the last couple of weeks has been token, uh, spend and how companies are going over budget and they're really trying to wrestle with the costs of am I actually seeing return on investment for the AM I'm spending on AI. And I'm curious what you're seeing in some of the enterprise startups that you back.

Speaker A: Yes, well, obviously, like, favorite topic. Right? Um, everyone was like, token maxing. Token maxing, Token maxing and, you know, everything else.

Speaker C: What did you think when that happened? Were you like, yeah, token maxing. Were you like, I don't know, maybe. Chill.

Speaker A: Yeah, I know, maybe. But then also I'm like, I'm thinking, like, do I token Max? Like, should I be using Claude more? You know what I mean? And then you think, like, it's hard

Speaker C: not to get hypey.

Speaker A: Totally. Yeah. I'm like, I should be using great. It's like, how can I use it? It's a good way of thinking, how can I use AI to automate more things? I do think. Then obviously some crazy headlines came out around Token maxing around again. Reportedly Uber kind of blowing through their annual budget in a few months. You had certain companies, like, shutting off Claude licenses for parts of the organization. I think Meta took away their leaderboard. And there's like another great quote around this, which I should have looked up. It's something like when a measure becomes a target, it no longer works as a meas. Right, right. Where it's just like, okay, let me figure out how I can run an infinite loop to max my, you know, AI usage. Um, so I feel like we went from token maxing to roi, which is the right, you know, right measure, I think, especially too, since it's like, okay, we're deploying agents everywhere. Theoretically, we're going to start deploying even more agents, so how can we actually Measure A, the real accurate spend per agent and per use case, per team member, per user, whatever that like unit is. And then B, how do we figure out the value against that once we've actually like accurately identified the cost? Um, and I do think that's the right question. I feel like that's what everyone's talking about now. And then you say, okay, well how do you actually do that? And I feel like we've now seen a few different approaches to that. Um, you know, all of which are really interesting and you know, probably some combination of these end up, you know, winning in the end. You've got a couple of standalone companies that are, you know, pretty new, I would say like very nascent kind of seed stage companies that are standalone applications working to um, do this, exactly this, you know, meter and track your AI usage, um, down to both, you know, unit, team use case, etc. Both from like tokens but like the whole built in cost, you know, where is the model being served? What uh, are the costs associated with serving at, what are the costs associated with the guardrails, you know, with, with maybe vision, etc. Um, um, so kind of like the all in cost there, those are the standalone plays. Then you had, you know, a couple of announcements which I think happened on the same day. So one of our portfolio companies, factory, announced uh, their model router. And so, you know, then you could say, okay, so maybe ROI is happening actually kind of at the harness level, at the harness layer.

Speaker C: Right. So the model router being like, so, so is it just being like, okay, I know what you're going to, what this task is and you currently have it on like Opus 4.8 where like a Sonnet 4.6 can handle this.

Speaker A: Yeah. So pick the best model for your particular, uh, you know, optimize in terms of like value and cost for your particular uh, desired outcome. And like kind of like let factory then do the routing. Right. So like optimizing cost in real time, if you will, uh, or optimizing ROI in real time. So there's like the harness layer, uh, argument. Right. And approach. And I do think that that, that makes a lot of sense. And then I'm pretty sure if not the same day or maybe you know, somewhere around that uh, you know, ramp amazing company and does a lot around you know, cost management spend I always think of as like the organization's like central nervous system for finances. You know, they also announced like basically like tokenomics, uh, but token spend and you know, intelligence around that. Right, right. And they had Eric had this, a great blog post or Twitter post X I should say, like intelligence is a third pillar.

Speaker C: Yeah, no, they did. They, we just, um, again at the time of this recording, I just, you know, published a little thing about one of their ramp index reports and they were saying that they were actually providing some interesting because, you know, Nvidia and Merkor recently were saying, you know, we're spending more on AI than humans. And it's like, is that what's happening? So it turns out like, not yet. And I don't know what, tell me what you think of this number. So it said the top 1% of firms, most like AI pilled firms, are spending 7,500 per employee per month on AI. Do you think that's, you know, and when we take this to like a monthly salary for an engineer, for a software engineer, it's around 1600. So it's a little less than half of what you would pay for an engineer. Does that sound like a reasonable number to you or.

Speaker A: It does. I definitely don't, I definitely don't think people are spending more on this than employees yet even like on a per employee basis. If you put it, you know, if you put it like that. I feel like honestly, kind of intuition wise, I feel like we're not even close to that yet.

Speaker C: Well, most companies aren't. Right? Most companies are somewhere around the like the top 10% spend only 611 per employee per month. Right. And then like everyone else, the median is like literally just paying the price of a seat, like 12 bucks.

Speaker A: Totally. Yeah. I would say like overall I feel like we're only at like 1 to 2% of um, AI adoption. Right? Across, across everyone. You know, if you look at like the top 1%, your numbers, you know, actually seem like I would believe them. Right. For the numbers you just gave and you know, can I see a world where all of a sudden that, you know, balloons maybe. But I actually think it still requires a lot of work to use AI. Well.

Speaker C: Oh yeah, yeah.

Speaker A: And especially now if you've got companies looking at roi, like if you think about it, you know, if you and I were, if we were literally tasked with how do I replace myself or how do I kind of like augment, how do I double myself or you know, 10x myself or 100x myself using uh, AI, I would like have to kind of sit down and map out, you know, what are my biggest, um, you know, a, like time intensive tasks, what are my best, like, kind of value drivers, how do I create AI systems? Like Train my own AI essentially. Right, obviously using some of the models and some of the applications. But like how do I map out my workflows that are most important? And it takes a decent amount of time to do that. Yeah, no one wants like, queries are one thing. No. And you really need to like, you know, kind of get these workflows done end to end. And so I still feel like, you know, not that there's friction there, but it requires some creative tinkering and you've got like these great people who are creative tinkering and figuring out like, how do I build myself in an AI native way? How do I build my entire team, my entire organization there? I feel like it tends to come from the founder down, um, on that stuff. But my guess is like even the Most, you know, AI pilled organizations, like we're still probably operating south of 50% and there's still so much room to grow.

Speaker C: One other thing about this report that I found interesting, it said the top 1% of firms tend to mix and match, opting to bounce between multiple frontier models and platforms that might give them cheaper access to open source models. Are you finding that? Or when you look at some of your portfolio companies, do they just have a Claude or an OpenAI subscription or are they using like Amazon Bedrock so that they can access all of them or how do they do it?

Speaker A: Yeah, I would honestly say it's pretty much a mix right now. And it really depends like based on team, uh, type. Right. So obviously in some cases like team members are just getting Claude. Right. Like it's like, okay, great, here's your, or whatever, let's say 90% of what they use on an AI, um, basis it's like maybe some version of Cloud or uh, or Perplexity or um, chatgpt. And then obviously on the engineering side, maybe you have Codex. Actually, I've heard that some teams are actually using Codex, you know, and cloud code, you know, and maybe uh, xyz, other tools as well. And then you also have like a bunch of applications or you know, companies like Factory, which you know, maybe you would put kind of in the, in the harness bucket. But you know, you kind of like there's some overlap between all these companies, but I think in many cases you actually have enterprises buying a lot of them. Um, and maybe we're still a little bit in the test and see mode right where it's. And maybe like I love using Perplexity but someone else loves Claude. Right. Or vice versa and you know, kind of like figuring out where the tools settle. Mhm. And so I think a little bit we're, you know, we're kind of in the, we're still testing, still in experimentation mode.

Speaker C: Yeah. Okay, so then going quickly back to the um, these kinds of like startups like Ramp, that you said and Factory, that are providing these services of, you know, managing your ROI, managing your spend, doing model routing, etc. Is this a real startup opportunity? Is this like a long term venture bet? Or are you as an investor looking at these kinds of startups that are offering these services and being like great M and a target?

Speaker A: Yeah, I think there's probably definitely a standalone path for, for some of the ROI companies. Like you know, maybe more of an observability play where you'd actually say like, you know, you don't necessarily want the application, you kind of want a third party to be, you know, to be tracking the cost and the roi. Uh, right. Versus like the model companies themselves because in some ways then it's like, you know, they're both delivering and measuring. So I guess there's like an observability layer that could live on top of that. Um, and you know, it could move from cost, just cost to like cost plus performance or you know, a variety of other things. It may be eventually it's like your agent control plane where you kind of can see all the different agents running, you know, all of your different use cases. More of like a. Um, I've also heard other people talk about as like the enterprise coo or like your AI coo, right. Where it's kind of like just monitoring, not just all the, you know, the employees you have, but like also all the agents you have running. Um, you know, I also think there's like the interesting angle of like an AI native workday, um, or an AI native HCM platform which is managing again, you know, as more and more agents are running. Uh, it's managing both employees and agents. So I think there's probably a standalone path, like exactly what it ends up looking like, I think a little bit tbd just since it's so early. Um, but I would say like, you know, ROI could be a good wedge. Um, but then I also think you really will see that same application or that same kind of, you know, end result accrue. Like I think Ramp will most likely be very successful at tracking this. Right. Because everyone's using it, uh, for everything. Right. For all of their costs. So it just, it makes so much natural sense. And I think too for Factory, like that model routing product makes so much natural sense. So my guess is there are going to be so many people or so many companies and opportunities where like an roi, like product will make sense, slot it in next to their other core products. Um, and so you'll probably end up have this like a view from a few different angles, I would guess.

Speaker C: Okay. And then in terms of, you know, helping with adoption and learning how to better use your tools, you know, we're seeing OpenAI, anthropic, et cetera, sending humans directly into enterprise organizations to sit alongside the users, figure out how they're automating specific workflows. These are called forward deployed engineers. When you and I discussed this originally, you called them a trojan horse. And I want to know why.

Speaker A: Yes. You know, obviously, like FDA is probably one of the most, like a favorite acronym right now. Uh, you know, I've now heard it like, referred to as like the forward deployed fill in the blank. Right. Because really everyone is. Or a lot of AI applied AI companies are thinking about a forward deployed motion, right? Forward deployed bankers, forward deployed lawyers, forward deployed fill in the blank. And you know, the reason I, you know, kind of like think of it as a trojan horse as it's a wonderful way to, uh, do a couple of things simultaneously like a, help your customers be very successful with AI adoption and using your product. So let's say you send in your fte, they kind of sit alongside the team at your customer, figure out how the customer is using your product, actually help them build these workflows that we were just talking about, right? Actually use the product effectively. And at the same time they're doing that. So like, okay, check mark on, let's get quick time to value at the customer. Let's really get them using our product. Well, um, so like, great, that's happening at the same time that's happening. They're actually then being able to see all of the other, um, gaps or holes and like where AI needs to exist. Right? So like, okay, let's use an example of, you know, you're sending in a forward deployed lawyer. You see all the end to end workflows the lawyer is working on. You know, your product today may be addressing 20% of that, um, but that FDE is now seeing like exactly where the product stops, exactly where your current, you know, product stops, exactly where it needs to go. It can kind, they can kind of map like, what are the, you know, highest value workflows either in terms of like repetition or value or um, you know, highest ROI workflows, essentially reporting those back to the, you know, the product team and then you know, especially given how quickly, you know, development cycle you

Speaker C: can just turn it over. Yeah, totally.

Speaker A: It's like, okay, tomorrow, you know, forward deploy, lawyer goes back in and there's like three new products that magically do exactly, you know, what the, what the customer needed. So I actually think like the FTEs are a really amazing way. They're truly like the um, you know, the workforce that like goes in and helps get AI adoption company like a

Speaker C: consultant on in a way, but way cooler. Yeah. Okay, two questions for you on this before we wrap up. Is this something that engineers are getting anxiety about? Are they reacting to this? Do they feel like this is complimenting or replacing their workflows?

Speaker A: Yeah, no, I actually think like people are getting really excited because if you think about it, you know, I actually think there's a real opportunity. I think it goes back to like the kind of how do you feel about AI and like are, you know, where are you on the doom scale of like, do you think it's really going to 10 or 100x people or you know, do you think it's going to replace people? Um, I would say like maybe as I'm a glass half full person, like I actually like tend to err on the side it's going to 10x or 100x people. And so you know, for me if I'm. Let's just like pretend there's a company that like is literally trying to AI for VCs. There's not yet, but eventually I'm sure there will be like AI for vcs. If I had someone, if I had like a forward deployed VC that came into my office and sat next to me and saw exactly what I was doing and the workflows that I was trying to automate and that person actually like helped me automate everything, I would be so appreciative. Right. I would love that. Um, because it's like, it's driving value for me faster. It's obviously then also driving value for the AIVC company and I just kind of think it's a win win. Um, because again like. But you have to believe ultimately end of the day there's like, there's probably like something that human time is still more valuable at.

Speaker C: Right.

Speaker A: You know, an adventure. Obviously it's like the face to face personal communication. So I'd love to have AI automate everything else. Right. And like save all my time for you know, meeting people and face to face interactions or talking to you.

Speaker C: Yeah, right. I'm curious if you think that, you know, the, the forward deployed engineer, I mean, they're sitting there, they're watching. Is there like an AI FDE that's watching the FDEs? And then we'll then take over their job and then you could just download some software to tell you where you need to add in some AI workflows.

Speaker A: Well, funny you say that, because people are clearly working on that. There's a lot of companies now. I would. I don't even know, like, what to call these companies. I kind of call them like, you know, basically like AI. They're trying to create, like, AI driven SOPs, um, but they actually literally, like, sit on your computer effectively is the way to think about them. They watch what you're doing. It was like a digital fte. They're gonna, like, see what you're doing within the or. And then they say like, oh, this was never codified before, but, like, here's your, you know, standard operating procedure, um, around, like, how this organization operates. So it tries to capture all this, uh, this data or knowledge that's like, ingrained only at the human level. Um, you know, I, I personally, I mean, I'm sure that will work too.

Speaker C: I think my cyber security editor just threw his laptop out the window.

Speaker A: Exactly, exactly. But, yeah, going back to our earliest conversation around security, um, you know, that has all those issues around it. So I don't know, I kind of like the FTE model right now. I think it works really well. Um, you think it's great? For so many reasons. So who knows, you know, will they both, Will they be at odds? Will they both win? I'm sure it's some combination of both.

Speaker C: Yeah. Okay. Well, hey, this has been super fun. I really appreciate you coming on to chat to us. Um, where can our listeners connect with you online if they want to?

Speaker A: Okay, amazing. Um, I would say LinkedIn. I'm, um, definitely on LinkedIn a lot. Any respond to, like, most of my, um, inmails and then I'm also on so Tiffany Luck on LinkedIn and then UCK TM on X. Cool. Didn't call Twitter this time.

Speaker C: Nice. Well done. Oh, uh, wow. You respond to most of your inmates. That's great. I don't do that.

Speaker A: I do for everyone. I do set up an agent for that. Rebecca just set up an agent.

Speaker C: I can't do it, so I don't want to train it anyway. Okay, well, thanks again for joining us on the show.

Speaker A: This was so fun. Okay, thanks for having me. I really enjoyed it.

Speaker C: And to our listeners, you can find me on LinkedIn too. Um, and I'm on Twitter X. I'm on all the, all the platforms and you can find Equity on threads and Twitter xquitypod. Uh, talk to you next time. Equity is hosted by TechCrunch senior reporters and produced by Teresa Loconsolo with editing by Cal. Subscribe on YouTube or wherever you get your podcasts and find out what's next@techcrunch.com events. Thanks so much for listening and we'll talk to you next time.