The Exchanges

Q Can you do that now? Like if I, if I wanted to implement, I mean, that sounds awesome. If I want to go and do that right now, how would I set that up?

A So you, you can do it today. It just requires a little bit of scripting in Airtable. And that itself you can actually generate with AI. So the idea is like you would go and create like a list of different people you want to send the video to, or even a list of topics you want to create a video for. And then we already have an Airtable, a scripting layer where you can define any arbitrary code. It's hosted in our infrastructure. And in this case, you would just define the code to go and talk to the HeyGen API. A lot of these AI products have APIs. And so you can go and create your own kind of orchestration layer and Airtable To go and automatically for each row ask Heijen to generate a video program programmatically for you using this avatar, and then you get a link to a Heijen video that you could then go and directly send to a customer or download it and then attach it, etc.

AI assessment note: “So you, you can do it today. It just requires a little bit of scripting”

Q Are you guys using any of this at Airtable?

A We actually do have an internal setup where it's all powered by Airtable. We use our own Airtable web agents to go in and research different, uh, target customers. We've learned a lot about them, and then we can actually generate customized email pitches for each of them. So right now it's just email, but we are experimenting with With more of the video stuff, we actually have a setup internally where we can now generate an avatar video powered by HeyGen automatically from content in a row of Airtable. So it could be a list of different event attendees, and that use case I just described, like, I may actually do it for, um, uh, thank you notes for, for different events that I host in the future. Like, if I host a, uh, a dinner with, with, um, our, uh, you know, some of our customers, I could actually now go and automate Creating a video that looks like it's me. And I would probably be upfront about it and say like, Hey, like here's an avatar version of how we like saying thank you for coming. And like the novelty of it itself actually would tickle people's fancy, but, um, but like to send a personalized thank you video from me to every attendee for all of our events.

AI assessment note: “We actually do have an internal setup where it's all powered by Airtable.”

Q Is that the framework for how you think of what's interesting for starting companies?

A I think so. I mean, I think there's, like, you can either go really broad, um, like, in this, you know, Cursor example or Cursor for X. So pick a market that's just so big, like there's so many potential target users, like tens of millions at least, and for whom the value is significant enough that they will pay for it. It's not a freemium or it's not purely like a free ad-driven product experience. It's not a consumer product. Pick something that's like broad enough and deep enough where there's just a very large pie to capture value out of, and then Even if you end up only capturing a niche of that pie, like you have a really large business on your hands. I think the alternate path is go really deep and narrow. So this is more like the property management example where it's like, you're probably never going to have like a hundred million users who need property management software. But like, if you go into that vertical and you attack it, you might have 5000. Yeah.

AI assessment note: “I think so. I mean, I think there's, like, you can either go really broad”

Q No, so the young guys nowadays, they just say, ChatGPT said this, and the doctor is going to say, well, I agree with you, and according to my belief, ChatGPT says blank, and ChatGPT, I think, is going to be the stamp of approval, or AI. Like, I think they're going to trust AI more than a human being. Does that make sense?

A Oh, for sure. I mean, I think, like, today, it's already better to, I mean, everybody should be taking all of their, you know, personal, like, uh, health diagnostics, you get a blood test or whatever, and, like, running that through ChatGPT, like, upload it, Asko three, like, Hey, like, here's my, my, uh, other symptoms. Here's how I feel. Like, tell me what I might have. Right. Or tell me what could be wrong. Like, um, I've, I've seen so many stories from like both personal friends, as well as like on Reddit, Twitter, et cetera, of like people who caught diagnoses that their doctor just missed, not necessarily because they're bad doctors. Although a lot of doctors probably are bad doctors, right?

AI assessment note: “Oh, for sure. I mean, I think, like, today, it's already better”

Q But these, these AI call companies, they're, they're still horrible. Are you bullish on them?

A Well, you know, I think the, um, the models have to get better at real-time synchronous, uh, interactions for, for, uh, You know, for, especially for voice calls, to fully get taken over by AI. I think for chat, like, you're chatting with somebody, like, it's already better in, in many cases, right? Because, like, you know, those chat experiences are horrible with humans, because, like, the human is probably doing, like, 20 different parallel chats, they don't really know or care about you, whereas an AI can be fully attentive, but I agree, I think the voice, uh, is just not quite there yet. Um, you've probably seen that Sesame, um, voice demo, where it's like, uh, uh, You know, one of the better real-time voice interactions.

AI assessment note: “I agree, I think the voice, uh, is just not quite there yet.”

Q Oh yes, I did see this. I absolutely, you think this is a good product?

A I, you know, it's a research team, and this is a demo product, but I think it shows where the models are going, and I think, you know, it could be from this team and this company, but also, I mean, you know that OpenAI, Um, and 11 Labs, et cetera, they're all working on making that interactivity better. OpenAI revealed the, the, the last voice mode release, where you could actually interrupt the AI halfway through a conversation, like, that was kind of a big deal. It's still not there yet, because it doesn't feel right. I mean, if you just talk to it, like, it doesn't feel like a human, but I think that will be, um, a frontier that we just, we pass or across in the next year or two.

AI assessment note: “this is a demo product, but I think it shows where the models are going”