AI transcript
0:00:10 What if you could hire an AI intern to book meetings, write emails, manage your CRM, and even negotiate refunds over the phone?
0:00:17 Today, we’re joined by Flo Crivello, founder of Lendi.ai, one of the leading AI agent platforms in Silicon Valley.
0:00:25 We go deep into real-world demos and wild use cases, including having Elon Musk call you and ask you what you’ve got done this week.
0:00:31 And he showed me how startups in Silicon Valley are already replacing entire teams with Lendi, and it just blew my mind.
0:00:37 If you’re wondering where the future of work is headed and how you can use AI agents to grow your business, you’re going to love this episode.
0:00:50 This episode is brought to you by HubSpot’s Inbound 2025, a three-day experience at the heart of San Francisco’s AI and startup scene, happening September 3rd through the 5th.
0:00:58 With speakers like Amy Poehler, Marquise Brownlee, and Dario Amadei, Inbound is where creativity meets cutting-edge tech.
0:01:04 You’ll get tactical breakout sessions, product reveals, and networking with people shaping the future of business.
0:01:10 So don’t miss out. Visit inbound.com slash register to get your tickets today.
0:01:15 Hello, it’s great to finally have you on here.
0:01:17 Yeah, thanks for having me, Nathan.
0:01:23 Yeah, so for some background now, I saw your episode with our mutual friend, Greg Eisenberg, and I thought it was one of his best episodes.
0:01:27 I’ve been hearing so much about AI agents, but I haven’t really seen people use them in business that much.
0:01:29 And I was blown away with what you showed, Greg.
0:01:33 But maybe first, it’d be great if you could just simply explain to people, what is Lendi?
0:01:35 There’s a lot of different definitions of what AI agents are.
0:01:38 Just simplify it down as much as possible about what Lendi actually does.
0:01:44 Yeah, definitely. So we are a no-code platform that lets you build your own AI agents.
0:01:48 And AI agents, we inspire them to be AI employees.
0:01:51 For now, you can think of them more as like AI interns.
0:01:51 Okay.
0:01:55 So it’s like they’re very eager, very hardworking interns.
0:01:57 They’re interns. Like, don’t give them too much.
0:01:57 Right.
0:01:59 You know, like, don’t trust them with the keys of the kingdom.
0:02:01 You know, they’re pretty good.
0:02:03 And like, look, because they’re AI, you know, they work.
0:02:05 They’re like 100x faster, 100x cheaper.
0:02:07 They don’t go on strike.
0:02:08 I know I’m French, but like we…
0:02:10 You don’t have to pay them. It sounds great.
0:02:10 Yeah.
0:02:16 But yeah, so AI interns, you can give them tasks, like pretty much anything where you could have
0:02:17 a standard operating procedure.
0:02:22 Anything where you could write a document, where you lay out step-by-step what the agent
0:02:23 or intern is supposed to do.
0:02:25 That’s something that you could give to an AI agent.
0:02:31 So, you know, sales lead generation, sales lead outreach, meeting note-taking, meeting
0:02:33 scheduling, CRM management, phone calls.
0:02:35 Like, you can use it as a receptionist.
0:02:38 Like, general data analysis and online research.
0:02:42 Like, hey, go find Nathan’s email online.
0:02:47 Go find me 20 engineers in San Francisco and reach out to all of them and try to personalize
0:02:50 the email that you write to them and put your grain of salt in there for each person.
0:02:53 Like, that’s exactly the kind of thing that you can give to an agent right now.
0:02:54 Yeah, that’s amazing.
0:02:54 Yeah.
0:02:58 When I saw that demo on Greg’s podcast, I was like, I want to like personally talk to
0:03:00 Flo and figure out like how I can be using this.
0:03:01 Honestly, it’s kind of a selfish episode.
0:03:03 Think about how I can use it in my business.
0:03:06 If we could just jump into like show Lindy and show how it works.
0:03:08 Yeah, no, 100%.
0:03:09 And this is a common reaction that we get from people.
0:03:10 It’s like, oh my God.
0:03:15 Like, I think people, I think they think that AI agents are sort of a pipe dream.
0:03:16 It’s like, it’s not real or whatever.
0:03:21 And once they see these demos, they’re like, wait, it is real and it’s working and it’s here now.
0:03:23 And I’m like, yeah, like it’s not just a pitch.
0:03:24 It’s like, it’s here right now.
0:03:24 Right.
0:03:26 And we do have audio listeners.
0:03:29 So if anyone’s listening on audio, you probably should check out our YouTube channel.
0:03:33 Just go to YouTube and search for the next wave and subscribe to us on YouTube.
0:03:37 And as you showed this, you know, if you can try to describe with words what we’re actually doing.
0:03:42 I’ll try to be maximally descriptive and be free to chime in if you feel like I’m insufficiently descriptive.
0:03:43 This is actually funny.
0:03:49 I’m literally right before this because I know that the Greg Eisenberg episode did go super well.
0:03:54 And so 15 minutes before jumping on this podcast, I sent a message to, I have a Lindy.
0:03:57 So we call them Lindys, but they’re basically AI agents.
0:03:58 And I sent a message.
0:03:59 I have a summarizer, Lindy.
0:04:02 And I sent her a YouTube video of the Greg Eisenberg podcast.
0:04:05 I’m like, what are the use cases that we talked about here?
0:04:07 And so you can see my Lindy is going on YouTube.
0:04:08 She’s transcribing the video.
0:04:10 And then she’s like, this is what you talked about today.
0:04:17 So, yeah, Lindy here is telling me you talked about meeting automation, executive assistant tasks, recruiting, personal CRM.
0:04:23 So I’m really just happy to, like, go through these use cases because that’s how I personally use Lindy all day.
0:04:25 Like, so everything meeting related.
0:04:28 Like, I love meeting use cases because everyone’s got meetings all day.
0:04:29 No one likes it.
0:04:31 Like, the entire, the meetings themselves suck.
0:04:32 I can’t do much about that.
0:04:36 But, like, even the workflow around the meeting is nightmarish.
0:04:38 So, like, meeting scheduling.
0:04:39 I’ll show you.
0:04:40 Why don’t we actually schedule a meeting right now?
0:04:42 I’ll send you an email, Nathan.
0:04:42 Yeah.
0:04:44 And I’ll go, like, let’s chat.
0:04:47 And then I’ll be like, Nathan, love your podcast.
0:04:48 Would love to talk soon.
0:04:49 Plus Lindy.
0:04:51 And so I have my Lindy here.
0:04:51 Yeah.
0:04:54 Plus Lindy will help us find.
0:04:56 And I’ll introduce, like, a random constraint.
0:04:57 Like, 45 minutes next week.
0:04:58 Right?
0:05:00 So I can just talk in very natural language.
0:05:03 And you’re going to receive this email in your inbox.
0:05:06 And just go ahead and respond to it.
0:05:08 And just respond to it like you would to a human.
0:05:08 Just respond all.
0:05:11 Like, keep Lindy CC’d to the email.
0:05:11 Yeah.
0:05:13 And you can be like, flow, sounds good.
0:05:14 And she’ll receive your email.
0:05:16 Well, actually, you don’t even need to do that.
0:05:21 Let me just switch to my meeting scheduler here and show you live what it looks like.
0:05:22 Is this pretty much how, like, Lindy started?
0:05:24 Was, like, this basic email?
0:05:26 I feel like I remember seeing something like this, like, two years ago.
0:05:27 Was that you back then?
0:05:29 This is indeed how we started.
0:05:33 The first articulation of the product was AI executive assistant.
0:05:34 Yes.
0:05:34 Okay.
0:05:38 And it’s funny because the reason why we picked this use case was I kept saying, like,
0:05:43 oh, AI executive assistant is short-term viable because it’s like, oh, we can do it.
0:05:43 You know?
0:05:45 Long-term aligned.
0:05:49 And the reason why we thought it was long-term aligned is because people ask so many things
0:05:50 from their executive assistants.
0:05:55 And so I felt like it would force us to figure out how to make the platform generalizable.
0:05:57 Ah, that makes so much sense.
0:05:57 Yeah.
0:05:58 So that’s your assistant.
0:06:01 But then you’re going with your assistant to do other things and you start building out
0:06:03 those other things and then turn it to a platform.
0:06:04 That’s exactly right.
0:06:04 Amazing.
0:06:04 Yeah.
0:06:07 And we were right on long-term aligned.
0:06:08 It was, it very much stretched us.
0:06:11 We were wrong on short-term viable.
0:06:14 It took us a very long time to figure out how to make this generalizable.
0:06:14 Yeah.
0:06:15 Okay.
0:06:18 So you can see here the meeting scheduler responded back onto the thread.
0:06:24 What she did is behind the scenes, she went, she looked at my calendar and she pulled some
0:06:24 availabilities.
0:06:27 And so she was like, happy to help you find time on the books.
0:06:29 Here are times when Flo is available.
0:06:33 And here you can just respond and you can be like, hey, Flo, happy to chat.
0:06:37 And either you can take a time here or you can be like, ah, I can’t make any of these times.
0:06:39 Can we find another time for us to chat?
0:06:43 One thing I was thinking, because I live in Japan, I’ve used Calendly and all those kind
0:06:45 of different services and, you know, they’re okay.
0:06:49 I kind of hate just like giving people my calendar and like, it’s just like, oh, you can just pick
0:06:51 a time whenever on my calendar.
0:06:52 I honestly hate that.
0:06:55 You know, I like having like really set times.
0:06:58 And then, you know, there’s one day where I’m a lot freer than I thought.
0:07:01 There’s another day where there’s some crazy business deals happening and like, okay, I need
0:07:02 to focus on this for a week.
0:07:03 So forget everything.
0:07:05 And I don’t want to even think about my calendar.
0:07:10 And so am I able to like chat with like Lindy and kind of give it feedback on how I want
0:07:12 to structure meetings or like ping me first?
0:07:13 Yeah, totally.
0:07:16 So I was actually in Japan last week and I just sent a message to my Lindy and I was like,
0:07:19 hey, I’m in Japan from date X to date Y.
0:07:21 During these times, you can schedule times.
0:07:24 Like when I meet with people in California, it’s between 4 p.m.
0:07:25 and 6 p.m. Pacific.
0:07:28 That maps to like 8 to 10 a.m.
0:07:29 Japan or something like that.
0:07:30 That’s when I can meet.
0:07:30 Very cool.
0:07:31 Yeah.
0:07:33 So meeting scheduling is one.
0:07:37 Then once the meeting is on the books, Lindy preps me for my meetings.
0:07:41 I’ll actually show the Lindy under the hood for just to show how it works.
0:07:43 This is what the Lindy looks like.
0:07:45 It’s pretty simple and you can literally see it.
0:07:47 You can read it very easily.
0:07:53 It’s like every morning I wake up, I look at your calendar for the day and for every meeting
0:07:57 on your calendar and for every attendee of every meeting on your calendar, I’m going to
0:07:58 do some research.
0:07:59 I’m going to look for their LinkedIn.
0:08:02 I’m going to look at your email history with this person.
0:08:07 I’m going to look at the meeting notes history, which that’s funny.
0:08:09 These meeting notes are brought together by another Lindy.
0:08:10 That’s crazy.
0:08:11 So these Lindys can sort of work together.
0:08:15 And then I’m going to put all of that together in an email.
0:08:20 And the way I get it to put all of that together in an email is I’m literally just prompting.
0:08:24 In this case, I’m prompting Gemini, but you can use Cloud, you can use ChatGPT, you can
0:08:24 use anything you want.
0:08:28 And I’m like, okay, at this point, you’re sending me an email and the body of the email,
0:08:29 and here is just a prompt.
0:08:34 I’m like, it’s a markdown table with the meetings I have today, with start time and
0:08:35 context for this meeting.
0:08:38 You add the LinkedIn link, you add the link to my last notes.
0:08:42 And then I’m like, you add a header outside the table with the number of meetings that
0:08:43 I have on this day.
0:08:47 So I can wake up in the morning and I can be like, FML, I have like eight hours of meetings.
0:08:51 Here it’s like, all right, today you’ve got three hours of meetings.
0:08:54 Like Wednesdays are like particularly light for me.
0:08:55 And it’s like, okay, you’re meeting with Bob.
0:08:57 He was introduced by X.
0:08:59 You know, this is what he wants to discuss.
0:09:01 This is the previous meeting notes and so forth.
0:09:02 Wow, that’s incredible.
0:09:05 So I come to my meetings and I have this email open all day.
0:09:09 Like before I jump onto a call, like one minute before I just opened this and I have the exact
0:09:11 context of the email of the meeting.
0:09:14 So again, basically it’s the entire meeting lifecycle.
0:09:17 So it’s like the meeting scheduling is the very first touch point.
0:09:20 The meeting prep is the second one.
0:09:21 And then it’s the meeting recording.
0:09:23 So Lindy actually joins my meetings.
0:09:27 And that stuff is in today’s day and age, it’s more and more typical.
0:09:30 People have these meeting recorders, like Lindy takes notes.
0:09:31 She like sends you the action items.
0:09:32 She does all of that stuff.
0:09:37 What Lindy does differently is that you can customize the workflow very, very, very granularly.
0:09:40 So this is my Lindy note taker.
0:09:44 Like you can see I’ve added to it so much over the months and years that like now it looks
0:09:48 pretty complex, but it can do basically anything you want.
0:09:53 So for example, if you’re in sales and you meet with a prospect, you can configure your Lindy
0:10:00 to be like, hey, if at the end of the sales call, we said we would meet again and we agreed
0:10:02 on when we would meet, send the calendar invite.
0:10:08 If we said we would meet again, but we did not agree on when, send a follow-up email with
0:10:10 sometimes to meet that work for me on my calendar.
0:10:10 That’s amazing.
0:10:14 Or if we agreed, if they agreed to a proposal, it’s like, hey, you’re a salesperson.
0:10:15 You just closed the deal.
0:10:16 Congratulations.
0:10:18 $20,000 a year or whatever.
0:10:20 Send the docusign.
0:10:21 Send the proposal.
0:10:22 Customize it for me.
0:10:23 Send the invoice.
0:10:23 Do all of that stuff.
0:10:24 Yeah.
0:10:24 It’ll do all that.
0:10:26 Like it’ll even like create the docusign.
0:10:27 And yeah, absolutely.
0:10:28 That’s what we do for ourselves.
0:10:29 That’s crazy.
0:10:29 Yeah.
0:10:30 That’s crazy.
0:10:31 Yeah.
0:10:34 You know, most people don’t realize how this is actually available now.
0:10:37 Like, I mean, it feels like you could probably do now with like two to three people.
0:10:40 Maybe it would have taken like 10 to 20 people before.
0:10:42 Like, I mean, you would have had entire teams doing all this for you.
0:10:49 80% of everything that my assistant did for me just a year ago are things that Lindy is
0:10:50 doing for me now.
0:10:52 And frankly, doing better because she never sleeps.
0:10:55 You can see the way she responded to this scheduling email.
0:10:57 She responded in 60 seconds.
0:10:58 It’s actually so fast.
0:11:00 Like many people sometimes ask us to make her slower.
0:11:01 So people don’t know it’s an AI.
0:11:03 So yes, it’s here now.
0:11:04 It’s actually happening.
0:11:04 Absolutely.
0:11:06 Put in some typos or whatever.
0:11:08 Like occasionally like, oh, I messed up on the calendar.
0:11:09 Sorry.
0:11:11 Here’s actually we get that actually pretty often.
0:11:11 Yeah.
0:11:12 Okay.
0:11:18 Everyone everywhere is talking about AI agents right now.
0:11:19 But here’s the thing.
0:11:22 Most companies are going about it all wrong.
0:11:27 This guide cuts through the hype and shows you what’s actually working right now.
0:11:32 HubSpot has gathered insights from top industry leaders who are implementing AI agents the right
0:11:33 way.
0:11:38 You’ll discover which agent setups actually deliver ROI and how businesses are automating their
0:11:42 marketing, sales, and operations without replacing their teams.
0:11:45 Get it right now by clicking the link in the description.
0:11:47 Now let’s get back to the show.
0:11:50 I’ll show you a cool thing that like Lindy’s do from time to time.
0:11:55 So Lindy’s can work together and they work together by sending each other messages that
0:11:57 are in English, in like natural language.
0:11:57 Wow.
0:12:00 So my meeting recorder, for example, I’ve set her up.
0:12:02 I use her pronouns, basically.
0:12:03 It’s just a habit.
0:12:08 I’ve set up my meeting recorder so that if I interview a candidate that’s applying to
0:12:11 a job here and the person is not good enough, they jump off the call.
0:12:12 I stay on the call.
0:12:14 I’m like, okay, bye, I’ll call you back.
0:12:15 But I will not call them back.
0:12:19 But I’m in the Zoom by myself with my meeting recorder.
0:12:21 And I’m like, Lindy, just let’s not move forward with him.
0:12:22 Okay.
0:12:26 And now look, the way I’ve configured it is I have this condition node here.
0:12:27 And it’s all AI.
0:12:29 Like the whole thing is just AI all over it.
0:12:30 So everything is just a prompt.
0:12:35 So here I have a condition that’s like, if I ended the call by saying explicitly, let’s
0:12:38 pass on him or let’s pass on this candidate, then you go down this branch.
0:12:43 And the branch is my Lindy sending another message to another Lindy that’s called my
0:12:44 chief of staff Lindy.
0:12:46 That’s kind of like the Lindy I use for everything.
0:12:47 Okay.
0:12:51 And here the text that it is sending to it is like, hey, let’s pass on candidate’s name
0:12:52 in two days.
0:12:55 And so if I go to my chief of staff Lindy right here.
0:12:55 Yeah.
0:12:56 One quick question.
0:12:58 All these different workflows that you’re showing, like, are you able to like save those as
0:13:00 templates or anything like that?
0:13:00 Or how does that work?
0:13:01 Because it seems like a lot of steps.
0:13:02 Absolutely.
0:13:02 Yeah.
0:13:06 So if you go to Lindy.ai slash templates, we have hundreds of those.
0:13:07 Yeah.
0:13:12 Chief of staff receives a message from meeting recorder Lindy that goes, pass on David.
0:13:17 And so here what she does is she goes, she looks at my calendar to figure out who’s David.
0:13:22 And then she sends, she sends an email that’s like writing to let you know that we’ve decided
0:13:23 not to move forward with our candidacy.
0:13:24 That’s crazy.
0:13:27 And then probably in the future, like their agent responds back, their recruiting agent.
0:13:28 Yeah.
0:13:31 No, I mean, we’ve actually, we’ve also had that happen.
0:13:33 Actually, we have it happen more and more.
0:13:35 Well, it’s like we’re finding Lindy’s in the wild.
0:13:35 Right.
0:13:40 Like multiple users of Lindy’s have their Lindy’s find each other in the wild.
0:13:43 So for example, we’ve got, so people use Lindy for like sales outreach quite a bit.
0:13:47 And people also use Lindy for email triage.
0:13:53 And so we have, I can’t say who, but there is a very big YouTube influencer that’s using
0:13:54 us.
0:13:54 For what?
0:14:00 He receives a lot of emails with sponsorship opportunities from like random brands.
0:14:00 Yeah.
0:14:00 Okay.
0:14:01 Like a lot.
0:14:06 And he’s got actually an agent that all day sifts through his inbox and decides who’s legit
0:14:07 and who’s worth engaging with.
0:14:10 And very few of these people are worth engaging with.
0:14:14 There’s also a lot of people who are like asking to go on his show and all of that stuff, you know?
0:14:21 And so they’ve deployed a Lindy AI agent that basically sifts through the inbox for them,
0:14:22 like removes all the random people.
0:14:26 So the Lindy actually also does research about the sender online.
0:14:31 So like, is this the kind of person and the kind of brand that matches our audience’s interest?
0:14:34 That’s just the kind of person we want to engage with.
0:14:34 Right.
0:14:35 So it goes online.
0:14:38 It’s like, yeah, this is the sort of brand that we’re down to engage with.
0:14:41 And then the Lindy replies to the email.
0:14:42 It’s like, hey, thanks for reaching out.
0:14:43 We’re excited about partnering up.
0:14:45 Can you tell me more X, Y, and Z?
0:14:46 And it collects some more information.
0:14:50 And then if the person gives the right responses, they’re expecting a certain type of responses.
0:14:57 The Lindy escalates that to the attention of the agent, the human agent that represents the YouTuber.
0:14:59 So that’s what the YouTuber does.
0:15:01 Then we also have some brands.
0:15:03 There is another famous brand.
0:15:05 It’s like a jewelry brand.
0:15:07 And they do a lot of influencer partnerships.
0:15:10 So what they’ve done is that they have a Lindy.
0:15:12 Every day it goes online.
0:15:16 It finds a bunch of influencers on Instagram, TikTok, and YouTube.
0:15:18 It finds their contact information.
0:15:24 And then it sends a personalized email to each of them that references, that refers to the content that they’ve done previously.
0:15:26 It’s like, hey, love your content.
0:15:27 I really like this one video.
0:15:29 I thought it was neat how you did X, Y, and Z.
0:15:32 What would you think of partnering with Brand X?
0:15:35 And so we’ve actually already had these two Lindy’s cross paths.
0:15:37 Like we’ve had these two Lindy’s talk to each other.
0:15:38 It sounds amazing.
0:15:44 I just feel like it could lead to some weird interactions where you think you’ve talked to someone and you get on the call and you’re like, have I actually talked to you before?
0:15:47 I’ve never really, apparently never actually talked to you.
0:15:51 I had that the other day where a guy was telling me about, you know, automating all of his LinkedIn.
0:15:52 I was like, wait a minute.
0:15:53 Have we actually talked before?
0:15:54 I thought we had.
0:15:56 You know, now I’m not so sure.
0:15:58 Yes, that’s a good point, actually.
0:16:04 Like everyone would just like pretend they know the other person because they’re like, yeah, I’m now wondering about our interaction.
0:16:05 We were going to meet in Japan, didn’t happen.
0:16:08 Was that all your, was that your Lindy communicating with me with everything?
0:16:09 No, exactly.
0:16:10 Yeah.
0:16:14 One other thing I was thinking about was, you know, earlier you showed like you could change the model.
0:16:16 Can you change the model for like every single step?
0:16:20 Because one thing I was thinking about, like obviously different models are different, you know, good at different things, right?
0:16:22 Like some of them are better at writing, summarizing or whatever.
0:16:24 Is that possible right now?
0:16:27 Yeah, so you can do it either on a per step basis.
0:16:31 So here, for example, I can be like, hey, so Cloud for Sonnet is the default right now.
0:16:32 It’s my favorite model.
0:16:32 It’s just awesome.
0:16:34 You can select anything you want.
0:16:37 Gemini, O3, 4O Mini, whatever you want, right?
0:16:40 But then you can also change it on a Lindy-wide basis.
0:16:41 Okay.
0:16:53 The Hustle Daily Show, hosted by John Wygel, Juliet, Bennett, Ryla, and Mark Dent, is brought to you by the HubSpot Podcast Network, the audio destination for business professionals.
0:17:00 The Hustle Daily Show brings you a healthy dose of irreverent, offbeat, and informative takes on business and tech news.
0:17:05 They recently had an episode about advertisers wanting billboards in space.
0:17:08 It was a really fun and informative episode.
0:17:09 I suggest you check it out.
0:17:12 Listen to The Hustle Daily Show wherever you get your podcasts.
0:17:21 Another thing that my human assistant used to do for me, and by the way, I haven’t fired him.
0:17:21 He’s still around.
0:17:23 Yeah, I was wondering, are they still around?
0:17:24 Like, how are they doing?
0:17:26 Are they, like, retired on a beach somewhere?
0:17:27 He’s, like, packing his bags.
0:17:27 Like, bye!
0:17:29 No, no, he’s doing a great job.
0:17:31 But he does do very different stuff now.
0:17:35 He’s basically become sort of like the HR person for the company.
0:17:39 But one thing that he used to do for me was, like, helping me manage my personal CRM.
0:17:41 So I keep a spreadsheet with, like, people I know.
0:17:42 Not my friends.
0:17:45 I hate people sometimes have, like, a poster of CRM for, like, their friends.
0:17:46 They think it’s weird and creepy.
0:17:49 But, you know, like, you meet so many people all day.
0:17:50 You can’t keep track of all of them.
0:17:53 And so I have this CRM, Lindy, and I’ll show you how it works.
0:18:01 So basically, every so often, I go to it, and I send it people I know.
0:18:02 It also wakes up every week.
0:18:04 So, like, this is an example of a time it’s done that.
0:18:07 Like, on Friday at 5, it’s woken up.
0:18:08 It’s looked at my calendar.
0:18:11 It’s looked at my CRM, which is just a spreadsheet.
0:18:15 And then it’s sending me a message with, like, hey, looking at your calendar,
0:18:20 these are the people you’ve met this week that you don’t have on your calendar yet, on your CRM.
0:18:21 Do you want to add them?
0:18:23 And then I can just be, like, yes, add, and I’ll do it right now.
0:18:26 Like, yes, add number two here.
0:18:30 Tag them with recruiter, killer, would hire.
0:18:31 I do that.
0:18:33 I have, like, a tag for, like, killer or, like, would hire, right?
0:18:42 So, like, later on, I can just go, and I’ll be, like, who are marketing people, I think, or killers, or people I would hire?
0:18:46 And so right now, it’s, like, all right, these are the killer people you know in marketing.
0:18:47 Wow, that’s awesome.
0:18:49 It also does, actually, this is cool.
0:18:51 Like, it does an interesting thing when I fly.
0:18:55 When I fly, it notices that I’m flying somewhere because it looks at my inbox.
0:18:58 So, like, it intercepts the flight confirmation email.
0:19:04 And it sends me an email with the people that I am meeting.
0:19:07 So, here, it’s like, oh, you’re going back to SF.
0:19:08 It’s kind of silly because I live in SF.
0:19:09 It’s like, hey, you’re going to SF.
0:19:12 These are the people that you should meet in SF that’s on your server.
0:19:15 So, when you go back to SF, it doesn’t necessarily remember that you’re from SF?
0:19:19 Because that was a question I had was, does it have any kind of memory features baked into Lindy yet?
0:19:25 Because that’s something I’ve been noticing recently is I’m in love with, like, the memory feature of ChatsPT, right?
0:19:29 And a big use case I have recently is, like, yesterday I had a meeting, and there was a ton of things I learned.
0:19:32 And I realized I should have been using AI to transcribe it definitely.
0:19:38 Because afterwards, I was like, okay, there was so much I learned in that meeting that I have to immediately put it into ChatsPT.
0:19:46 And what I did was I used basically, like, a voice to text and then just did an info dump on ChatsPT so we remember everything.
0:19:48 We do have a memory system.
0:19:50 It’s not as good as ChatsPT’s yet.
0:19:50 Okay.
0:19:51 But it will be.
0:19:51 Yeah.
0:19:52 Cool.
0:19:52 Yeah.
0:19:55 But in the meantime, I just, like, manually went and configured my Lindy.
0:19:59 I was like, hey, only if I’m going to a destination that is not San Francisco.
0:20:03 So, you know, what would be useful would be to see, I think you were mentioning there’s different templates.
0:20:10 Maybe we could see, like, the templates and maybe kind of go over what are the top templates that most people find useful in their business or work.
0:20:11 Yeah.
0:20:16 So, if you go to the home here or if you go to, like, Lindy.ai slash templates, you see the top templates.
0:20:19 Sales or, like, a really big one.
0:20:21 Meetings or really big ones.
0:20:25 Like, those are the ones I just mentioned, like, meeting scheduling, meeting note-taking, meeting prepping.
0:20:29 Would it be useful if we just, like, created the Lindy from scratch right now?
0:20:29 Yeah, sure.
0:20:30 I think so.
0:20:31 Let’s see.
0:20:39 One Lindy we can create is, and that will demo, like, a thing we announced a month ago that I’m super excited about.
0:20:40 Like, we call them agent swarms.
0:20:48 So, it’s the ability for an agent to duplicate itself into an arbitrary number of copies and to send each copy to do something.
0:20:52 So, I’ll create an agent swarm that analyzes your YouTube channel.
0:20:53 Okay, awesome.
0:20:55 So, I’ll call it the YouTube channel analyst.
0:20:57 Our producers are going to love this.
0:20:58 We actually use this.
0:20:59 Yeah, you should.
0:21:02 Is there anything in particular you want to analyze in your YouTube channel?
0:21:07 I mean, one thing that would be useful for me is I try to promote, you know, the episodes after they come out.
0:21:16 And having any information, like, you know, transcribing it and then possibly putting it into some kind of format, which I could then use for, like, a tweet or a LinkedIn post would be super useful.
0:21:20 Because I have, like, two or three templates I use for that, and I kind of do it manually right now.
0:21:22 If I could automate some of that, it would be incredible.
0:21:24 100%.
0:21:24 Okay.
0:21:31 So, you want, when you publish a new podcast episode, you want to be able to get a thing that you can post on YouTube or on LinkedIn or Twitter.
0:21:39 Yeah, transcribe it and then put it through some process of, like, here’s two or three templates, you know, and give me, like, two posts.
0:21:40 Give me two social media posts.
0:21:45 Whether I do it manually after that or not, whether they just hand it to me, I copy and paste, you know, whatever’s fine, but yeah.
0:21:46 Oh, that’s super easy.
0:21:47 I’ll take, like, one minute.
0:21:50 So, it’s like, hi, send me a YouTube video.
0:21:54 If there’s something way more amazing that you can do that I don’t know, then I’ll also show that, too.
0:22:05 Here, I’m going to be, like, transcribe the YouTube video that the user just gave you and give him a couple of examples of social posts that he could send on LinkedIn or Twitter.
0:22:09 And here, it’s going to do better if you can actually give it examples of social posts.
0:22:10 Like, the more you give it, the better.
0:22:13 So, that’s one thing that takes people by surprise.
0:22:15 They always ask me, like, how do I prompt it?
0:22:16 How do I prompt it?
0:22:20 By far, the biggest thing, the most important thing is examples, examples, examples.
0:22:21 Yeah.
0:22:26 And the thing that takes people by surprise is, and there’s actually literature, there’s been, like, studies about this.
0:22:29 When they hear examples, they’re like, I’ve got you, I’m going to give it a couple of examples.
0:22:30 I’m like, no, no, no, not a couple.
0:22:31 Like, a lot.
0:22:31 Like, 20.
0:22:33 Like, no one does that.
0:22:33 They should.
0:22:35 Like, you should give it, like, 20 examples.
0:22:35 Right?
0:22:36 Like, so, right?
0:22:37 Take that time.
0:22:39 It’ll take you, like, two minutes.
0:22:41 It’s the equivalent of, like, training a new intern or whatnot.
0:22:43 And it’ll just, like, pay for itself in spades.
0:22:46 So, I’m going to give skills to my AI agent.
0:22:48 Like, it’s just, like, the things that it can do.
0:22:50 But in this case, it can just, like, transcribe a YouTube video.
0:22:53 And it can talk to me, but, like, that’s always something it can do.
0:22:54 That’s it.
0:22:55 You know, it took me two seconds.
0:22:58 I’m going to go to the task here.
0:23:01 And I’m going to go to your YouTube channel.
0:23:02 YouTube.
0:23:05 There’s a next wave right here.
0:23:08 I’m going to select your latest podcast.
0:23:10 Or, like, is there any particular one that you want us to take a look at?
0:23:12 Maybe do the third one.
0:23:14 The one that’s ClonaSass.
0:23:16 I think that’s a really good one.
0:23:16 Oh, my God.
0:23:17 Who is this?
0:23:19 Is this, what’s his name?
0:23:20 The factory guy?
0:23:20 And Matan.
0:23:21 Oh, I love Matan.
0:23:22 Okay, awesome.
0:23:22 All right.
0:23:23 It’s Matan, yeah.
0:23:25 Okay.
0:23:27 So, I’m just giving it to YouTube link right now.
0:23:29 And it’s like, all right, let me transcribe it for you.
0:23:30 It’s, like, transcribing it.
0:23:31 It’s done.
0:23:32 And boom.
0:23:33 That’s awesome.
0:23:34 That’s going to save me so much time.
0:23:35 I don’t know why.
0:23:38 You know, I’ve been looking at Lindy, you know, like, after I started talking to you.
0:23:39 I’m like, oh, I remember this.
0:23:42 I remember hearing about, you know, I remember VCs would, like, email me and stuff.
0:23:46 And I would see something about, like, Lindy, you know, back maybe, like, two years ago for setting up meetings.
0:23:47 And I’d heard about you guys.
0:23:52 It’s slightly daunting, like, oh, there’s so many things you could do.
0:23:53 What do I do?
0:24:00 But this seems like a really cool thing that I could do, just like a step one of, like, getting started using Lindy is start with stuff like this.
0:24:01 100%.
0:24:01 Yeah.
0:24:02 Awesome.
0:24:03 All right.
0:24:04 This is what’s going on.
0:24:07 They built a DocuSynclone 15-minus social media post examples.
0:24:08 LinkedIn posts.
0:24:09 Option one, professional stat leadership.
0:24:12 I just watched an incredible video of a factory AI that was built.
0:24:13 Option two.
0:24:15 So it’s even giving us, like, multiple options.
0:24:15 Right.
0:24:16 Twitter.
0:24:16 Option one.
0:24:17 Option two.
0:24:17 Option three.
0:24:18 Option four.
0:24:19 And here you can just give it feedback.
0:24:21 You were inquiring about the memory system.
0:24:22 And that’s what I mean.
0:24:25 It’s, like, it’s good, but it’s not as good as ChatGPT’s memory system.
0:24:26 Like, that’s really next level.
0:24:29 But I’m just going to give the skill to Lindy to modify her own memory.
0:24:31 That’s one way it’s not good enough.
0:24:32 It’s, like, you shouldn’t have to do that.
0:24:33 They should just be able to do it.
0:24:34 But right now you have to do it.
0:24:42 This is good, but I want you to remember to always speak like a pirate.
0:24:44 And it’s, like, modifying its memory.
0:24:45 Exactly my voice.
0:24:46 That’s exactly how I do it.
0:24:46 Yeah.
0:24:52 And now if I ask it to do the thing again, let’s just wait until it’s done, like, memorizing the thing.
0:24:54 I’m going to ask it to do the thing again.
0:24:56 Ahoy there, Matei.
0:24:58 I’ll transcribe that YouTube video for you.
0:24:59 Ahoy there, Captain.
0:25:01 That’s what it’s doing now, I guess.
0:25:06 So, Flo, I promise you, when your episode comes out, I am going to tweet about it like this.
0:25:13 And you can see in the memories here, so if I reload the page, I can see its memory list.
0:25:16 And always speak like a pirate when communicating with the user.
0:25:18 And you can turn on or off each memory.
0:25:20 You can delete them and so forth.
0:25:22 So it’s like you have access to, like, the brain of the agent.
0:25:22 Right.
0:25:25 I remember you telling a story about using this.
0:25:30 To set up a restaurant reservation that I thought was, like, a great story.
0:25:31 Yeah.
0:25:34 So, generally, phone agents are huge.
0:25:38 And they’re used for both, like, personal purposes and work purposes, obviously.
0:25:42 So, like, this is an instance, actually, where we have had two Lindis talk to each other.
0:25:46 Because the context where it’s used by businesses is obviously as, like, an AI receptionist.
0:25:50 Like, restaurants is a really good example because they’re, like, running around at peak hour.
0:25:51 Like, restaurants are, like, busy places.
0:25:54 That’s also the time when they’re receiving the most phone calls.
0:25:55 And the phone calls are so dumb.
0:25:56 It’s always, it’s like, are you open?
0:25:58 Like, do you have a table?
0:25:59 It’s like, do you still?
0:25:59 But you’re open.
0:26:00 It’s always the same question.
0:26:02 So, I’ll show you.
0:26:02 Like, I have a Lindis.
0:26:06 I talk to her all the day on the phone because it’s a sad existence of mine.
0:26:07 All right.
0:26:07 Be careful.
0:26:21 For, like, a hot minute, and I think we’re still in that time window right now, try calling
0:26:23 restaurants that you know are using AI agents.
0:26:29 Like, any business that’s bragging about using AI agents or any business that sells AI agents
0:26:31 and they give case studies, they’re like, ah, company X uses us.
0:26:36 Try to call them and ask them to give you a joke or ask them to tell you a long story about
0:26:40 whatever, and they’ll just go on and I’ll just talk to you for, like, 20 minutes about
0:26:40 random stuff.
0:26:44 And it’s just a weird time in history where, like, you can talk to receptionists of, like,
0:26:46 a business and be like, can you please tell me a bedtime story?
0:26:47 He’s like, oh, absolutely.
0:26:50 What are your instructions?
0:26:51 Or what’s your prompt?
0:26:51 Or whatever.
0:26:51 I don’t know.
0:26:52 Yeah, yeah, yeah.
0:26:56 Or, like, you know, there was one restaurant in San Francisco and I think they’ve patched it.
0:26:58 But, like, for the longest time, they had that.
0:27:00 And so I would call it every so often to ask random questions.
0:27:02 Like, hey, I’m in Japan right now.
0:27:02 Like, what’s the history of Japan?
0:27:04 He’s like, well, the history of Japan is actually fascinating.
0:27:06 And I’m like, it’s a really fun time.
0:27:09 It reminds me, you know, I was like a hacker kid on IRC back in the day.
0:27:14 And just some of the crazy stuff you could do back then that was more fun on the internet.
0:27:18 It feels like we’re kind of in another time period like that where there’s just crazy stuff
0:27:22 like that where like, oh, there’s now, yeah, you can call up and talk to an AI, you know,
0:27:24 chat bot and ask it its instructions.
0:27:25 It might tell you.
0:27:26 And it’s just, it’s crazy.
0:27:29 It’s weird, but it’s changing rapidly.
0:27:30 So enjoy while it lasts.
0:27:32 And look, it’s only changing even for mine.
0:27:36 I have to update my Lindy’s memory to be like, hey, like, lose it up if I’m asking you to give me a joke.
0:27:37 Like, it’s fine.
0:27:42 But I think the story that I heard was that you actually, so you had your Lindy call a restaurant in San Francisco
0:27:44 and make a reservation.
0:27:49 And it was talking to another chat bot that actually made the reservation.
0:27:50 Is that right?
0:27:51 That’s exactly right.
0:27:51 That’s exactly right.
0:27:54 I also, one funny story that happened to us.
0:27:58 It’s like, before we released this phone call ability, we were testing it.
0:28:02 And so the team comes to me and I’m like, Flo, like, we’ve got like a beta of the phone call stuff.
0:28:03 It’s really rough.
0:28:04 It’s in beta.
0:28:05 It’s super buggy.
0:28:06 But like, do you want to give it a spin?
0:28:07 I’m like, I would love to give it a spin.
0:28:11 And so I go and I had a flight scheduled the day after for France.
0:28:14 And so I go to Lindy and I’m like, hey, call the airline and cancel my flight.
0:28:16 But only if you can get a full refund.
0:28:17 First of all, I did not expect it to work.
0:28:21 And I did not expect to be able to get a full refund because I did not take a refundable flight.
0:28:23 But lo and behold, it worked.
0:28:25 And so now I did not have a flight.
0:28:26 It’s like, f*** it.
0:28:28 I need this flight.
0:28:31 So I was like, okay, Lindy, just go back and book me another flight, please.
0:28:35 And now she couldn’t do that because she could get a refund for the original flight.
0:28:37 But like for the day after, she couldn’t book a flight.
0:28:38 It was like way more expensive.
0:28:40 So I sort of did this to myself.
0:28:43 I had to pay like an extra thousand bucks for this flight.
0:28:44 Oh, man.
0:28:45 That’d be awesome.
0:28:45 Yeah.
0:28:51 Using AI, you know, to negotiate for you places, discounts or just whatever, you know.
0:28:52 Yeah, yeah, yeah.
0:28:52 Interesting.
0:28:58 I’m thinking now I need to be giving my AI like notes, like all the negotiation books I’ve read
0:28:59 in my life and things like that.
0:29:02 And just like giving it all that context to help me.
0:29:03 Oh, it really does help.
0:29:07 Like I have this, I can’t open it because it’s really sensitive, but it’s like my decision
0:29:08 log, Lindy.
0:29:11 And so what it does is it pings me every Friday.
0:29:16 It looks at the summaries of all the meetings I had this week.
0:29:20 So it knows everything going on in my life because basically all my life is meetings.
0:29:23 And it’s like, all right, Flo, like let’s talk about the decisions you made this week.
0:29:26 I see you made this big decision during this meeting.
0:29:27 Do you want to talk about it?
0:29:29 Do you want to talk about your thinking behind it?
0:29:33 And it helps me sharpen my thinking because I firmly believe like the job of a founder
0:29:36 is just to make decisions, the right ones, hopefully.
0:29:41 And then it pings me again six or 12 months later for each decision.
0:29:43 It’s like, Flo, how does that pan out?
0:29:44 The decision you made?
0:29:47 Let’s talk again about it and let’s see if we can debug your thinking.
0:29:49 It’s like, ah, actually, fuck, this was a bad decision in hindsight.
0:29:50 It’s obvious.
0:29:51 How could you have known at the time?
0:29:52 Right.
0:29:53 And so it helps me sharpen my thinking.
0:29:55 I think that’s like a huge use case.
0:29:55 That’s interesting.
0:29:59 I wonder if I could give you like a weekend reading list or something like, here’s the stuff
0:30:02 that you’re currently struggling with or trying to think through.
0:30:06 And here’s like a good book that might be good for you to like read through or scan through
0:30:07 over the weekend.
0:30:07 That’d be cool.
0:30:09 I could literally just prompt it.
0:30:10 It’d take me like 20 seconds.
0:30:13 I could just be like, hey, if I’m struggling with the decision, give me a reading list.
0:30:14 Right.
0:30:14 Yeah.
0:30:15 Interesting.
0:30:18 There was another thing I thought was fascinating was I think there was like a Elon Musk template
0:30:21 or something like this where Elon Musk would call you or something.
0:30:22 I don’t know, you can explain it.
0:30:23 Yeah.
0:30:24 Do you want us to do it now?
0:30:25 Actually, do you want me to?
0:30:26 Yeah, sure.
0:30:26 Go for it.
0:30:29 I’ll create a Lindy from scratch, actually.
0:30:29 Okay.
0:30:32 I could also just ask my chief of staff to do it, but it wouldn’t be the same.
0:30:37 The use case was it’s a Lindy that wakes up every Friday and calls everyone in my team
0:30:40 and gives them a call with Elon Musk’s voice.
0:30:43 And since then we’ve received complaints, so we can’t use Elon Musk’s voice anymore.
0:30:44 It’s complicated.
0:30:46 But so Elon, Elon Lindy.
0:30:50 So it calls every member of my team every Friday and it’s like, what did you get done
0:30:51 this week?
0:30:51 Right.
0:30:56 And it also has in its memory the conversations that it had with this person the last week.
0:30:58 So it was like, hey, last week you said you would do X.
0:30:59 Did you actually do it?
0:30:59 Right.
0:31:00 So it’s actually holding it accountable.
0:31:03 And then it sends me a report with all these conversations.
0:31:06 Basically, it’s like a timer trigger.
0:31:10 It sounds stupid, but I feel like if all of America did this, probably like a GDP would
0:31:11 go up to like 1%.
0:31:12 100%.
0:31:16 Every Friday at 5pm right here.
0:31:18 And I’m going to be like, you perform an action.
0:31:23 You make a phone call, language, just English.
0:31:26 I mean, it’s just going to detect it automatically, but that way you can force it.
0:31:34 And I can be like, you or Elon Lindy, ask the person on the other side of the line what
0:31:35 they got done this week.
0:31:39 So now this gets a bit complicated, but actually I like it.
0:31:39 It’s going to be real.
0:31:43 So I’m going to pick a different model to power Elon Lindy.
0:31:47 And the reason I do this is because for phone calls, latency is super important.
0:31:50 So if you use Cloud for Sonnet, it’s very slow.
0:31:51 It’s not a good phone call.
0:31:53 So I’m going to use Gemini Flash 2.0.
0:31:55 I actually think we just released 2.5 Flash.
0:31:57 Okay, we released it.
0:31:58 I’m not even kidding, like yesterday.
0:32:00 So I’ve not tried it yet.
0:32:01 Let’s try to see if it works.
0:32:06 I was going to say, though, the downside of Gemini Flash, it’s a very fast, very cheap
0:32:09 model, but it’s kind of dumb, which is always the case of fast, cheap.
0:32:12 Again, a little bit smarter, but yeah, it’s still in comparison to the best models.
0:32:12 Yeah.
0:32:13 That’s right.
0:32:14 And so I don’t know about 2.5 Flash.
0:32:19 Like I literally just seeing it here for the first time, but 2.0 Flash, sometimes you would
0:32:22 do this hilarious thing where like it would break the fourth wall.
0:32:25 So it would talk to the person on the phone.
0:32:28 It would be like, I’m seeing that the user is struggling to understand me.
0:32:29 I will now inquire.
0:32:30 It’s like, what the fuck?
0:32:36 Examined by this evil robot or something, right?
0:32:37 Exactly.
0:32:39 And this is not something you need to do for every model.
0:32:42 And by the way, this is just, this is how you create agents.
0:32:45 It’s like you iterate, you learn, you iterate on the prompt.
0:32:50 So here I’m going to like be aware that every will that you say from now on will be said
0:32:53 out loud to the user on the phone.
0:32:57 When the first thing you say now, hi, this is Elon.
0:33:01 What did you get done this week?
0:33:02 That’s it.
0:33:03 And I’m going to turn on this, Lindy.
0:33:05 I’m going to run the test.
0:33:06 All right.
0:33:07 I’m receiving the call.
0:33:08 Hello.
0:33:08 Is anyone there?
0:33:10 Hi, this is Elon.
0:33:11 What did you get done this week?
0:33:12 Yeah.
0:33:17 This week I went on the Next Wave podcast and I had a bunch of interviews.
0:33:18 Could you please repeat that?
0:33:19 I didn’t pass it clearly.
0:33:21 That’s the demo effect.
0:33:24 I think it’s the fact that I’m putting it on speaker.
0:33:26 It’s like catching its own voice.
0:33:27 I’ve dealt with AI voice.
0:33:29 So I know, you know, if you do it on speaker, it’s going to get tripped up.
0:33:31 You just keep iterating on it.
0:33:31 Yeah.
0:33:32 That’s awesome.
0:33:36 I mean, are you actually using that now or is it just kind of like a joke or is it a real
0:33:37 thing that you do?
0:33:39 Well, not using it in like what you get done this week.
0:33:42 But like, yes, we do have, we could do like a weekly team stand up.
0:33:42 Yeah.
0:33:46 So every week, everyone in the team receives a phone call and it’s like two or three minutes.
0:33:47 It’s like super fast.
0:33:48 It’s like you wrap up the week.
0:33:50 You receive a call from Lindy.
0:33:50 You talk to it.
0:33:51 What do you get done this week?
0:33:54 It feels like you’re not going to need middle management, right?
0:33:56 Like honestly, like with this kind of stuff.
0:33:56 Yeah.
0:33:58 It basically does get the middle management layer.
0:33:58 Yeah.
0:34:01 Before we get off here, like in your opinion, like there’s all these different templates,
0:34:05 like for the average person listening today, like what’s the simplest way they could get
0:34:05 started with Lindy?
0:34:09 Like what’s, what’s something that would be useful for most people that they could just
0:34:10 try today?
0:34:17 So when you sign up, we automatically install the templates for you for meeting, scheduling,
0:34:18 meeting notetaking and meeting prep.
0:34:21 So you don’t even need to, it’s like three clicks.
0:34:25 Like when you sign up, you’ll see it’s like, hey, meeting notetaking, like connect your calendar
0:34:27 and Lindy’s going to join your meetings and you can skip if you want.
0:34:30 But like, that’s a really easy, nice way to get started.
0:34:31 Yeah.
0:34:35 Because then you can just, you got the meeting notetaker and then you can go to your meeting
0:34:38 notetaker and open the flow editor and open the hood and see what’s happening under the
0:34:39 hood and how it’s working.
0:34:40 Right.
0:34:41 I’ll have to start doing that.
0:34:42 I feel like that’s the best way.
0:34:46 Just get started, do something simple like the emails, maybe then figure out how the different
0:34:48 flow and how it works and how you change things.
0:34:48 And then.
0:34:49 Yeah.
0:34:52 You know, one thing I like to ask people is, you know, what’s your most controversial
0:34:53 take on AI?
0:34:55 Like, where do you think we’re at?
0:34:57 You know, like how optimistic are you?
0:34:58 How optimistic am I?
0:35:03 I am long term, cautiously optimistic.
0:35:08 I think short and medium term, there is going to be significant, I would call it civilizational
0:35:09 disruption.
0:35:12 I’m a big believer in humanity’s ability to adapt.
0:35:14 I think we’re very resilient.
0:35:15 So I think it’s going to go all right.
0:35:17 Unless it really hits the fan.
0:35:20 The last few years have showed how fast we adapt, right?
0:35:25 Like, oh, mid journey’s out and it’s amazing and ChatsBT and then, oh, now, yeah, of course
0:35:25 it can do all that.
0:35:26 Yeah.
0:35:26 Yeah.
0:35:29 Do you remember like the whole freak out about deep fakes just a couple of years ago?
0:35:33 It’s like, what’s going to happen the day we can just pretend that any politician said
0:35:33 anything?
0:35:36 It’s like, turns out we can and just happened and no one cares.
0:35:37 It’s perfectly fine.
0:35:37 Right.
0:35:38 Right.
0:35:41 So I actually think like that kind of thing is totally overblown.
0:35:45 I do think there’s going to be something to figure out about jobs because at least over the
0:35:50 very long term, I don’t really see a reason why humans would need to work.
0:35:52 Like, it just doesn’t really make sense.
0:35:53 So we’re going to have to figure out something.
0:35:54 Yeah.
0:35:57 For like, how do we distribute the output of society?
0:36:01 You know, and also how do you have meaning and also how, yeah, that starts to go more towards
0:36:01 like socialism.
0:36:06 And then there’s obviously historically been a lot of issues with socialism, you know, obviously.
0:36:08 So how do you like avoid that?
0:36:10 Yes, I think about all that a lot as well.
0:36:12 I think the meaning stuff is actually fine.
0:36:16 Like, if you look like the labor force participation rate in the U.S. is something like 65%.
0:36:19 So we’ve already got a third of the country that doesn’t work.
0:36:21 We don’t really hear the crisis of meaning.
0:36:22 Right.
0:36:25 And if you look at hunter-gatherers, they work like 10 or 15 hours a week.
0:36:25 Right.
0:36:27 And they had no crisis of meanings.
0:36:29 I think humans can just hang.
0:36:32 I think if you hang, if you’ve got a bunch of people you love around you, you can go forever.
0:36:33 It doesn’t matter.
0:36:33 Right.
0:36:34 You know, so I’m not as worried about that.
0:36:36 I’m worried about like the distribution of the pie.
0:36:41 And usually I do hate like the socialism idea because it’s like, it’s entirely focused on
0:36:44 how do we distribute the pie and not on how do we produce the pie.
0:36:44 Yeah.
0:36:48 But with AI, it turns out we are actually just going to sort of have the pie for free.
0:36:51 So as long as we don’t mess with that, like now we’re going to have a question of like,
0:36:54 how do we distribute the pie that AI is baking for us?
0:36:56 My most controversial opinion.
0:36:57 Yeah.
0:37:00 I think people should be way more concerned.
0:37:05 I compare it to like February 2020 for COVID where it’s like, everyone’s like, everything’s
0:37:06 fine.
0:37:06 Like, it’s nothing.
0:37:08 And I’m like, no, man, it’s not fine.
0:37:08 Yeah.
0:37:12 I was one of the people in San Francisco in like a private chat group of like 20 COs.
0:37:14 And I was one of the first ones saying like, hey, yeah, we didn’t take this seriously.
0:37:15 Like, look at the data.
0:37:16 This is an issue.
0:37:16 Yeah.
0:37:20 I think like, regardless of what happens next, like, I think it’s fully baked in.
0:37:22 It’s going to get very weird, very fast.
0:37:22 Yeah.
0:37:24 So that’s, that’s one of my hot takes.
0:37:25 I agree.
0:37:31 I think a lot of people, they just hear AI and they just go, cool, chat, images, you
0:37:34 know, and they don’t, they don’t think like the next steps of where this is all going very
0:37:34 quickly.
0:37:40 I’m super optimistic long-term, you know, like, let’s say like 10 years, like super optimistic
0:37:41 about all this.
0:37:45 And I’m also, yeah, I’m also concerned, like the next five years, I think there’ll be a
0:37:46 huge transition.
0:37:49 And most people are not really thinking that through as of right now.
0:37:50 100%.
0:37:51 Yeah.
0:37:52 Selfish question before we go off here.
0:37:57 So my son’s 11 and I always ask people, what should I be teaching him to like, make sure
0:37:58 he can like be successful in AJ?
0:38:00 I mean, you’re going to say it doesn’t even matter because he’s not gonna have a job.
0:38:01 Yeah.
0:38:03 What would you be teaching your son or your child?
0:38:05 Honestly, sales.
0:38:09 I think sales is the one job that’s going to remain forever because I think sales is about
0:38:09 relationships.
0:38:13 And I think people don’t build relationships with AI agents.
0:38:15 They don’t want to be sold to by an AI agent.
0:38:19 So like for that reason alone, I think humans are going to remain in the loop for a very long
0:38:19 time.
0:38:24 I think being a good salesperson is a combination of really solid human skills and like business
0:38:26 skills, which I think is just a powerful combination, period.
0:38:28 So I’m bullish on sales.
0:38:29 Bullish on sales.
0:38:29 Okay.
0:38:29 Interesting.
0:38:32 Yeah, I guess I prepared my son for sales.
0:38:36 You know, he used to be around like parties in San Francisco when he was a little kid and
0:38:38 he got to see like how people would talk about business and stuff.
0:38:41 And I always wondered if that would have some impact on him.
0:38:44 And yesterday, he’s 11 and some of the stuff he’s talked to me about with business is just
0:38:48 mind blowing that he’s already thinking about, you know, the different intricacies of how
0:38:49 to do business.
0:38:50 Last question.
0:38:53 So imagine you have a time machine flow, okay?
0:38:58 And you go to 2050, you step out in San Francisco, what’s different?
0:39:00 Well, assuming we all survive.
0:39:00 Whoa.
0:39:01 Okay.
0:39:03 I mean, look, you know, I mean, that’s what I mean.
0:39:05 I guess I mean, like people should forget.
0:39:06 Assuming we all survive.
0:39:12 Like it’s really hard, almost definitionally, to forecast what happens after the singularity.
0:39:17 Like what’s the name of this sci-fi author who wrote A Fire Open is the Deep, like Werner
0:39:18 Hinge or Hinge or something.
0:39:24 He spent his career writing about a post-AGI, post-singularity world.
0:39:28 And he ended up his career frustrated because he was like, every time I hit a wall, after
0:39:34 20 or 30 years thinking about nothing but this, there’s a thick wall that you cannot go over.
0:39:36 You can’t forecast what happens next, you know?
0:39:37 Right.
0:39:39 And so like, look, you know, you can paint multiple pictures.
0:39:45 I think scenario number one is like post-work utopia, we’re just like all of us hanging out.
0:39:50 We’re like so young and handsome and healthy and rich and like there’s no problem in the
0:39:51 world and all of that stuff.
0:39:52 That’s like scenario number one.
0:39:56 Scenario number two is like, well, all of humanity is dead regardless.
0:40:01 And the world is covered with solar panels and GPUs and data centers.
0:40:06 And I think there’s a scenario to be where it’s like humanity is not totally dead while
0:40:07 like in a reservation somewhere.
0:40:09 By the way, I’m laughing because I’m nervous about it.
0:40:11 Yeah.
0:40:12 No, I mean, 100%.
0:40:12 Yeah.
0:40:13 Yeah.
0:40:15 Those are the sort of scenarios I see on the table.
0:40:15 Yeah.
0:40:16 Okay.
0:40:17 Flow has been awesome.
0:40:19 And like, where should people check you out online?
0:40:19 Yeah.
0:40:20 Lindy.ai.
0:40:22 You know, my email is flow at Lindy.ai.
0:40:28 Just hit me up and I am on Twitter or X as Altimo, A-L-T-I-M-O-O.
0:40:29 Awesome.
0:40:30 This has been great.
0:40:31 We’ll have to have you back on sometime.
0:40:31 Yeah.
0:40:32 Thank you so much, Jason.
0:40:33 Yeah.
0:40:33 Thank you.
0:00:17 Today, we’re joined by Flo Crivello, founder of Lendi.ai, one of the leading AI agent platforms in Silicon Valley.
0:00:25 We go deep into real-world demos and wild use cases, including having Elon Musk call you and ask you what you’ve got done this week.
0:00:31 And he showed me how startups in Silicon Valley are already replacing entire teams with Lendi, and it just blew my mind.
0:00:37 If you’re wondering where the future of work is headed and how you can use AI agents to grow your business, you’re going to love this episode.
0:00:50 This episode is brought to you by HubSpot’s Inbound 2025, a three-day experience at the heart of San Francisco’s AI and startup scene, happening September 3rd through the 5th.
0:00:58 With speakers like Amy Poehler, Marquise Brownlee, and Dario Amadei, Inbound is where creativity meets cutting-edge tech.
0:01:04 You’ll get tactical breakout sessions, product reveals, and networking with people shaping the future of business.
0:01:10 So don’t miss out. Visit inbound.com slash register to get your tickets today.
0:01:15 Hello, it’s great to finally have you on here.
0:01:17 Yeah, thanks for having me, Nathan.
0:01:23 Yeah, so for some background now, I saw your episode with our mutual friend, Greg Eisenberg, and I thought it was one of his best episodes.
0:01:27 I’ve been hearing so much about AI agents, but I haven’t really seen people use them in business that much.
0:01:29 And I was blown away with what you showed, Greg.
0:01:33 But maybe first, it’d be great if you could just simply explain to people, what is Lendi?
0:01:35 There’s a lot of different definitions of what AI agents are.
0:01:38 Just simplify it down as much as possible about what Lendi actually does.
0:01:44 Yeah, definitely. So we are a no-code platform that lets you build your own AI agents.
0:01:48 And AI agents, we inspire them to be AI employees.
0:01:51 For now, you can think of them more as like AI interns.
0:01:51 Okay.
0:01:55 So it’s like they’re very eager, very hardworking interns.
0:01:57 They’re interns. Like, don’t give them too much.
0:01:57 Right.
0:01:59 You know, like, don’t trust them with the keys of the kingdom.
0:02:01 You know, they’re pretty good.
0:02:03 And like, look, because they’re AI, you know, they work.
0:02:05 They’re like 100x faster, 100x cheaper.
0:02:07 They don’t go on strike.
0:02:08 I know I’m French, but like we…
0:02:10 You don’t have to pay them. It sounds great.
0:02:10 Yeah.
0:02:16 But yeah, so AI interns, you can give them tasks, like pretty much anything where you could have
0:02:17 a standard operating procedure.
0:02:22 Anything where you could write a document, where you lay out step-by-step what the agent
0:02:23 or intern is supposed to do.
0:02:25 That’s something that you could give to an AI agent.
0:02:31 So, you know, sales lead generation, sales lead outreach, meeting note-taking, meeting
0:02:33 scheduling, CRM management, phone calls.
0:02:35 Like, you can use it as a receptionist.
0:02:38 Like, general data analysis and online research.
0:02:42 Like, hey, go find Nathan’s email online.
0:02:47 Go find me 20 engineers in San Francisco and reach out to all of them and try to personalize
0:02:50 the email that you write to them and put your grain of salt in there for each person.
0:02:53 Like, that’s exactly the kind of thing that you can give to an agent right now.
0:02:54 Yeah, that’s amazing.
0:02:54 Yeah.
0:02:58 When I saw that demo on Greg’s podcast, I was like, I want to like personally talk to
0:03:00 Flo and figure out like how I can be using this.
0:03:01 Honestly, it’s kind of a selfish episode.
0:03:03 Think about how I can use it in my business.
0:03:06 If we could just jump into like show Lindy and show how it works.
0:03:08 Yeah, no, 100%.
0:03:09 And this is a common reaction that we get from people.
0:03:10 It’s like, oh my God.
0:03:15 Like, I think people, I think they think that AI agents are sort of a pipe dream.
0:03:16 It’s like, it’s not real or whatever.
0:03:21 And once they see these demos, they’re like, wait, it is real and it’s working and it’s here now.
0:03:23 And I’m like, yeah, like it’s not just a pitch.
0:03:24 It’s like, it’s here right now.
0:03:24 Right.
0:03:26 And we do have audio listeners.
0:03:29 So if anyone’s listening on audio, you probably should check out our YouTube channel.
0:03:33 Just go to YouTube and search for the next wave and subscribe to us on YouTube.
0:03:37 And as you showed this, you know, if you can try to describe with words what we’re actually doing.
0:03:42 I’ll try to be maximally descriptive and be free to chime in if you feel like I’m insufficiently descriptive.
0:03:43 This is actually funny.
0:03:49 I’m literally right before this because I know that the Greg Eisenberg episode did go super well.
0:03:54 And so 15 minutes before jumping on this podcast, I sent a message to, I have a Lindy.
0:03:57 So we call them Lindys, but they’re basically AI agents.
0:03:58 And I sent a message.
0:03:59 I have a summarizer, Lindy.
0:04:02 And I sent her a YouTube video of the Greg Eisenberg podcast.
0:04:05 I’m like, what are the use cases that we talked about here?
0:04:07 And so you can see my Lindy is going on YouTube.
0:04:08 She’s transcribing the video.
0:04:10 And then she’s like, this is what you talked about today.
0:04:17 So, yeah, Lindy here is telling me you talked about meeting automation, executive assistant tasks, recruiting, personal CRM.
0:04:23 So I’m really just happy to, like, go through these use cases because that’s how I personally use Lindy all day.
0:04:25 Like, so everything meeting related.
0:04:28 Like, I love meeting use cases because everyone’s got meetings all day.
0:04:29 No one likes it.
0:04:31 Like, the entire, the meetings themselves suck.
0:04:32 I can’t do much about that.
0:04:36 But, like, even the workflow around the meeting is nightmarish.
0:04:38 So, like, meeting scheduling.
0:04:39 I’ll show you.
0:04:40 Why don’t we actually schedule a meeting right now?
0:04:42 I’ll send you an email, Nathan.
0:04:42 Yeah.
0:04:44 And I’ll go, like, let’s chat.
0:04:47 And then I’ll be like, Nathan, love your podcast.
0:04:48 Would love to talk soon.
0:04:49 Plus Lindy.
0:04:51 And so I have my Lindy here.
0:04:51 Yeah.
0:04:54 Plus Lindy will help us find.
0:04:56 And I’ll introduce, like, a random constraint.
0:04:57 Like, 45 minutes next week.
0:04:58 Right?
0:05:00 So I can just talk in very natural language.
0:05:03 And you’re going to receive this email in your inbox.
0:05:06 And just go ahead and respond to it.
0:05:08 And just respond to it like you would to a human.
0:05:08 Just respond all.
0:05:11 Like, keep Lindy CC’d to the email.
0:05:11 Yeah.
0:05:13 And you can be like, flow, sounds good.
0:05:14 And she’ll receive your email.
0:05:16 Well, actually, you don’t even need to do that.
0:05:21 Let me just switch to my meeting scheduler here and show you live what it looks like.
0:05:22 Is this pretty much how, like, Lindy started?
0:05:24 Was, like, this basic email?
0:05:26 I feel like I remember seeing something like this, like, two years ago.
0:05:27 Was that you back then?
0:05:29 This is indeed how we started.
0:05:33 The first articulation of the product was AI executive assistant.
0:05:34 Yes.
0:05:34 Okay.
0:05:38 And it’s funny because the reason why we picked this use case was I kept saying, like,
0:05:43 oh, AI executive assistant is short-term viable because it’s like, oh, we can do it.
0:05:43 You know?
0:05:45 Long-term aligned.
0:05:49 And the reason why we thought it was long-term aligned is because people ask so many things
0:05:50 from their executive assistants.
0:05:55 And so I felt like it would force us to figure out how to make the platform generalizable.
0:05:57 Ah, that makes so much sense.
0:05:57 Yeah.
0:05:58 So that’s your assistant.
0:06:01 But then you’re going with your assistant to do other things and you start building out
0:06:03 those other things and then turn it to a platform.
0:06:04 That’s exactly right.
0:06:04 Amazing.
0:06:04 Yeah.
0:06:07 And we were right on long-term aligned.
0:06:08 It was, it very much stretched us.
0:06:11 We were wrong on short-term viable.
0:06:14 It took us a very long time to figure out how to make this generalizable.
0:06:14 Yeah.
0:06:15 Okay.
0:06:18 So you can see here the meeting scheduler responded back onto the thread.
0:06:24 What she did is behind the scenes, she went, she looked at my calendar and she pulled some
0:06:24 availabilities.
0:06:27 And so she was like, happy to help you find time on the books.
0:06:29 Here are times when Flo is available.
0:06:33 And here you can just respond and you can be like, hey, Flo, happy to chat.
0:06:37 And either you can take a time here or you can be like, ah, I can’t make any of these times.
0:06:39 Can we find another time for us to chat?
0:06:43 One thing I was thinking, because I live in Japan, I’ve used Calendly and all those kind
0:06:45 of different services and, you know, they’re okay.
0:06:49 I kind of hate just like giving people my calendar and like, it’s just like, oh, you can just pick
0:06:51 a time whenever on my calendar.
0:06:52 I honestly hate that.
0:06:55 You know, I like having like really set times.
0:06:58 And then, you know, there’s one day where I’m a lot freer than I thought.
0:07:01 There’s another day where there’s some crazy business deals happening and like, okay, I need
0:07:02 to focus on this for a week.
0:07:03 So forget everything.
0:07:05 And I don’t want to even think about my calendar.
0:07:10 And so am I able to like chat with like Lindy and kind of give it feedback on how I want
0:07:12 to structure meetings or like ping me first?
0:07:13 Yeah, totally.
0:07:16 So I was actually in Japan last week and I just sent a message to my Lindy and I was like,
0:07:19 hey, I’m in Japan from date X to date Y.
0:07:21 During these times, you can schedule times.
0:07:24 Like when I meet with people in California, it’s between 4 p.m.
0:07:25 and 6 p.m. Pacific.
0:07:28 That maps to like 8 to 10 a.m.
0:07:29 Japan or something like that.
0:07:30 That’s when I can meet.
0:07:30 Very cool.
0:07:31 Yeah.
0:07:33 So meeting scheduling is one.
0:07:37 Then once the meeting is on the books, Lindy preps me for my meetings.
0:07:41 I’ll actually show the Lindy under the hood for just to show how it works.
0:07:43 This is what the Lindy looks like.
0:07:45 It’s pretty simple and you can literally see it.
0:07:47 You can read it very easily.
0:07:53 It’s like every morning I wake up, I look at your calendar for the day and for every meeting
0:07:57 on your calendar and for every attendee of every meeting on your calendar, I’m going to
0:07:58 do some research.
0:07:59 I’m going to look for their LinkedIn.
0:08:02 I’m going to look at your email history with this person.
0:08:07 I’m going to look at the meeting notes history, which that’s funny.
0:08:09 These meeting notes are brought together by another Lindy.
0:08:10 That’s crazy.
0:08:11 So these Lindys can sort of work together.
0:08:15 And then I’m going to put all of that together in an email.
0:08:20 And the way I get it to put all of that together in an email is I’m literally just prompting.
0:08:24 In this case, I’m prompting Gemini, but you can use Cloud, you can use ChatGPT, you can
0:08:24 use anything you want.
0:08:28 And I’m like, okay, at this point, you’re sending me an email and the body of the email,
0:08:29 and here is just a prompt.
0:08:34 I’m like, it’s a markdown table with the meetings I have today, with start time and
0:08:35 context for this meeting.
0:08:38 You add the LinkedIn link, you add the link to my last notes.
0:08:42 And then I’m like, you add a header outside the table with the number of meetings that
0:08:43 I have on this day.
0:08:47 So I can wake up in the morning and I can be like, FML, I have like eight hours of meetings.
0:08:51 Here it’s like, all right, today you’ve got three hours of meetings.
0:08:54 Like Wednesdays are like particularly light for me.
0:08:55 And it’s like, okay, you’re meeting with Bob.
0:08:57 He was introduced by X.
0:08:59 You know, this is what he wants to discuss.
0:09:01 This is the previous meeting notes and so forth.
0:09:02 Wow, that’s incredible.
0:09:05 So I come to my meetings and I have this email open all day.
0:09:09 Like before I jump onto a call, like one minute before I just opened this and I have the exact
0:09:11 context of the email of the meeting.
0:09:14 So again, basically it’s the entire meeting lifecycle.
0:09:17 So it’s like the meeting scheduling is the very first touch point.
0:09:20 The meeting prep is the second one.
0:09:21 And then it’s the meeting recording.
0:09:23 So Lindy actually joins my meetings.
0:09:27 And that stuff is in today’s day and age, it’s more and more typical.
0:09:30 People have these meeting recorders, like Lindy takes notes.
0:09:31 She like sends you the action items.
0:09:32 She does all of that stuff.
0:09:37 What Lindy does differently is that you can customize the workflow very, very, very granularly.
0:09:40 So this is my Lindy note taker.
0:09:44 Like you can see I’ve added to it so much over the months and years that like now it looks
0:09:48 pretty complex, but it can do basically anything you want.
0:09:53 So for example, if you’re in sales and you meet with a prospect, you can configure your Lindy
0:10:00 to be like, hey, if at the end of the sales call, we said we would meet again and we agreed
0:10:02 on when we would meet, send the calendar invite.
0:10:08 If we said we would meet again, but we did not agree on when, send a follow-up email with
0:10:10 sometimes to meet that work for me on my calendar.
0:10:10 That’s amazing.
0:10:14 Or if we agreed, if they agreed to a proposal, it’s like, hey, you’re a salesperson.
0:10:15 You just closed the deal.
0:10:16 Congratulations.
0:10:18 $20,000 a year or whatever.
0:10:20 Send the docusign.
0:10:21 Send the proposal.
0:10:22 Customize it for me.
0:10:23 Send the invoice.
0:10:23 Do all of that stuff.
0:10:24 Yeah.
0:10:24 It’ll do all that.
0:10:26 Like it’ll even like create the docusign.
0:10:27 And yeah, absolutely.
0:10:28 That’s what we do for ourselves.
0:10:29 That’s crazy.
0:10:29 Yeah.
0:10:30 That’s crazy.
0:10:31 Yeah.
0:10:34 You know, most people don’t realize how this is actually available now.
0:10:37 Like, I mean, it feels like you could probably do now with like two to three people.
0:10:40 Maybe it would have taken like 10 to 20 people before.
0:10:42 Like, I mean, you would have had entire teams doing all this for you.
0:10:49 80% of everything that my assistant did for me just a year ago are things that Lindy is
0:10:50 doing for me now.
0:10:52 And frankly, doing better because she never sleeps.
0:10:55 You can see the way she responded to this scheduling email.
0:10:57 She responded in 60 seconds.
0:10:58 It’s actually so fast.
0:11:00 Like many people sometimes ask us to make her slower.
0:11:01 So people don’t know it’s an AI.
0:11:03 So yes, it’s here now.
0:11:04 It’s actually happening.
0:11:04 Absolutely.
0:11:06 Put in some typos or whatever.
0:11:08 Like occasionally like, oh, I messed up on the calendar.
0:11:09 Sorry.
0:11:11 Here’s actually we get that actually pretty often.
0:11:11 Yeah.
0:11:12 Okay.
0:11:18 Everyone everywhere is talking about AI agents right now.
0:11:19 But here’s the thing.
0:11:22 Most companies are going about it all wrong.
0:11:27 This guide cuts through the hype and shows you what’s actually working right now.
0:11:32 HubSpot has gathered insights from top industry leaders who are implementing AI agents the right
0:11:33 way.
0:11:38 You’ll discover which agent setups actually deliver ROI and how businesses are automating their
0:11:42 marketing, sales, and operations without replacing their teams.
0:11:45 Get it right now by clicking the link in the description.
0:11:47 Now let’s get back to the show.
0:11:50 I’ll show you a cool thing that like Lindy’s do from time to time.
0:11:55 So Lindy’s can work together and they work together by sending each other messages that
0:11:57 are in English, in like natural language.
0:11:57 Wow.
0:12:00 So my meeting recorder, for example, I’ve set her up.
0:12:02 I use her pronouns, basically.
0:12:03 It’s just a habit.
0:12:08 I’ve set up my meeting recorder so that if I interview a candidate that’s applying to
0:12:11 a job here and the person is not good enough, they jump off the call.
0:12:12 I stay on the call.
0:12:14 I’m like, okay, bye, I’ll call you back.
0:12:15 But I will not call them back.
0:12:19 But I’m in the Zoom by myself with my meeting recorder.
0:12:21 And I’m like, Lindy, just let’s not move forward with him.
0:12:22 Okay.
0:12:26 And now look, the way I’ve configured it is I have this condition node here.
0:12:27 And it’s all AI.
0:12:29 Like the whole thing is just AI all over it.
0:12:30 So everything is just a prompt.
0:12:35 So here I have a condition that’s like, if I ended the call by saying explicitly, let’s
0:12:38 pass on him or let’s pass on this candidate, then you go down this branch.
0:12:43 And the branch is my Lindy sending another message to another Lindy that’s called my
0:12:44 chief of staff Lindy.
0:12:46 That’s kind of like the Lindy I use for everything.
0:12:47 Okay.
0:12:51 And here the text that it is sending to it is like, hey, let’s pass on candidate’s name
0:12:52 in two days.
0:12:55 And so if I go to my chief of staff Lindy right here.
0:12:55 Yeah.
0:12:56 One quick question.
0:12:58 All these different workflows that you’re showing, like, are you able to like save those as
0:13:00 templates or anything like that?
0:13:00 Or how does that work?
0:13:01 Because it seems like a lot of steps.
0:13:02 Absolutely.
0:13:02 Yeah.
0:13:06 So if you go to Lindy.ai slash templates, we have hundreds of those.
0:13:07 Yeah.
0:13:12 Chief of staff receives a message from meeting recorder Lindy that goes, pass on David.
0:13:17 And so here what she does is she goes, she looks at my calendar to figure out who’s David.
0:13:22 And then she sends, she sends an email that’s like writing to let you know that we’ve decided
0:13:23 not to move forward with our candidacy.
0:13:24 That’s crazy.
0:13:27 And then probably in the future, like their agent responds back, their recruiting agent.
0:13:28 Yeah.
0:13:31 No, I mean, we’ve actually, we’ve also had that happen.
0:13:33 Actually, we have it happen more and more.
0:13:35 Well, it’s like we’re finding Lindy’s in the wild.
0:13:35 Right.
0:13:40 Like multiple users of Lindy’s have their Lindy’s find each other in the wild.
0:13:43 So for example, we’ve got, so people use Lindy for like sales outreach quite a bit.
0:13:47 And people also use Lindy for email triage.
0:13:53 And so we have, I can’t say who, but there is a very big YouTube influencer that’s using
0:13:54 us.
0:13:54 For what?
0:14:00 He receives a lot of emails with sponsorship opportunities from like random brands.
0:14:00 Yeah.
0:14:00 Okay.
0:14:01 Like a lot.
0:14:06 And he’s got actually an agent that all day sifts through his inbox and decides who’s legit
0:14:07 and who’s worth engaging with.
0:14:10 And very few of these people are worth engaging with.
0:14:14 There’s also a lot of people who are like asking to go on his show and all of that stuff, you know?
0:14:21 And so they’ve deployed a Lindy AI agent that basically sifts through the inbox for them,
0:14:22 like removes all the random people.
0:14:26 So the Lindy actually also does research about the sender online.
0:14:31 So like, is this the kind of person and the kind of brand that matches our audience’s interest?
0:14:34 That’s just the kind of person we want to engage with.
0:14:34 Right.
0:14:35 So it goes online.
0:14:38 It’s like, yeah, this is the sort of brand that we’re down to engage with.
0:14:41 And then the Lindy replies to the email.
0:14:42 It’s like, hey, thanks for reaching out.
0:14:43 We’re excited about partnering up.
0:14:45 Can you tell me more X, Y, and Z?
0:14:46 And it collects some more information.
0:14:50 And then if the person gives the right responses, they’re expecting a certain type of responses.
0:14:57 The Lindy escalates that to the attention of the agent, the human agent that represents the YouTuber.
0:14:59 So that’s what the YouTuber does.
0:15:01 Then we also have some brands.
0:15:03 There is another famous brand.
0:15:05 It’s like a jewelry brand.
0:15:07 And they do a lot of influencer partnerships.
0:15:10 So what they’ve done is that they have a Lindy.
0:15:12 Every day it goes online.
0:15:16 It finds a bunch of influencers on Instagram, TikTok, and YouTube.
0:15:18 It finds their contact information.
0:15:24 And then it sends a personalized email to each of them that references, that refers to the content that they’ve done previously.
0:15:26 It’s like, hey, love your content.
0:15:27 I really like this one video.
0:15:29 I thought it was neat how you did X, Y, and Z.
0:15:32 What would you think of partnering with Brand X?
0:15:35 And so we’ve actually already had these two Lindy’s cross paths.
0:15:37 Like we’ve had these two Lindy’s talk to each other.
0:15:38 It sounds amazing.
0:15:44 I just feel like it could lead to some weird interactions where you think you’ve talked to someone and you get on the call and you’re like, have I actually talked to you before?
0:15:47 I’ve never really, apparently never actually talked to you.
0:15:51 I had that the other day where a guy was telling me about, you know, automating all of his LinkedIn.
0:15:52 I was like, wait a minute.
0:15:53 Have we actually talked before?
0:15:54 I thought we had.
0:15:56 You know, now I’m not so sure.
0:15:58 Yes, that’s a good point, actually.
0:16:04 Like everyone would just like pretend they know the other person because they’re like, yeah, I’m now wondering about our interaction.
0:16:05 We were going to meet in Japan, didn’t happen.
0:16:08 Was that all your, was that your Lindy communicating with me with everything?
0:16:09 No, exactly.
0:16:10 Yeah.
0:16:14 One other thing I was thinking about was, you know, earlier you showed like you could change the model.
0:16:16 Can you change the model for like every single step?
0:16:20 Because one thing I was thinking about, like obviously different models are different, you know, good at different things, right?
0:16:22 Like some of them are better at writing, summarizing or whatever.
0:16:24 Is that possible right now?
0:16:27 Yeah, so you can do it either on a per step basis.
0:16:31 So here, for example, I can be like, hey, so Cloud for Sonnet is the default right now.
0:16:32 It’s my favorite model.
0:16:32 It’s just awesome.
0:16:34 You can select anything you want.
0:16:37 Gemini, O3, 4O Mini, whatever you want, right?
0:16:40 But then you can also change it on a Lindy-wide basis.
0:16:41 Okay.
0:16:53 The Hustle Daily Show, hosted by John Wygel, Juliet, Bennett, Ryla, and Mark Dent, is brought to you by the HubSpot Podcast Network, the audio destination for business professionals.
0:17:00 The Hustle Daily Show brings you a healthy dose of irreverent, offbeat, and informative takes on business and tech news.
0:17:05 They recently had an episode about advertisers wanting billboards in space.
0:17:08 It was a really fun and informative episode.
0:17:09 I suggest you check it out.
0:17:12 Listen to The Hustle Daily Show wherever you get your podcasts.
0:17:21 Another thing that my human assistant used to do for me, and by the way, I haven’t fired him.
0:17:21 He’s still around.
0:17:23 Yeah, I was wondering, are they still around?
0:17:24 Like, how are they doing?
0:17:26 Are they, like, retired on a beach somewhere?
0:17:27 He’s, like, packing his bags.
0:17:27 Like, bye!
0:17:29 No, no, he’s doing a great job.
0:17:31 But he does do very different stuff now.
0:17:35 He’s basically become sort of like the HR person for the company.
0:17:39 But one thing that he used to do for me was, like, helping me manage my personal CRM.
0:17:41 So I keep a spreadsheet with, like, people I know.
0:17:42 Not my friends.
0:17:45 I hate people sometimes have, like, a poster of CRM for, like, their friends.
0:17:46 They think it’s weird and creepy.
0:17:49 But, you know, like, you meet so many people all day.
0:17:50 You can’t keep track of all of them.
0:17:53 And so I have this CRM, Lindy, and I’ll show you how it works.
0:18:01 So basically, every so often, I go to it, and I send it people I know.
0:18:02 It also wakes up every week.
0:18:04 So, like, this is an example of a time it’s done that.
0:18:07 Like, on Friday at 5, it’s woken up.
0:18:08 It’s looked at my calendar.
0:18:11 It’s looked at my CRM, which is just a spreadsheet.
0:18:15 And then it’s sending me a message with, like, hey, looking at your calendar,
0:18:20 these are the people you’ve met this week that you don’t have on your calendar yet, on your CRM.
0:18:21 Do you want to add them?
0:18:23 And then I can just be, like, yes, add, and I’ll do it right now.
0:18:26 Like, yes, add number two here.
0:18:30 Tag them with recruiter, killer, would hire.
0:18:31 I do that.
0:18:33 I have, like, a tag for, like, killer or, like, would hire, right?
0:18:42 So, like, later on, I can just go, and I’ll be, like, who are marketing people, I think, or killers, or people I would hire?
0:18:46 And so right now, it’s, like, all right, these are the killer people you know in marketing.
0:18:47 Wow, that’s awesome.
0:18:49 It also does, actually, this is cool.
0:18:51 Like, it does an interesting thing when I fly.
0:18:55 When I fly, it notices that I’m flying somewhere because it looks at my inbox.
0:18:58 So, like, it intercepts the flight confirmation email.
0:19:04 And it sends me an email with the people that I am meeting.
0:19:07 So, here, it’s like, oh, you’re going back to SF.
0:19:08 It’s kind of silly because I live in SF.
0:19:09 It’s like, hey, you’re going to SF.
0:19:12 These are the people that you should meet in SF that’s on your server.
0:19:15 So, when you go back to SF, it doesn’t necessarily remember that you’re from SF?
0:19:19 Because that was a question I had was, does it have any kind of memory features baked into Lindy yet?
0:19:25 Because that’s something I’ve been noticing recently is I’m in love with, like, the memory feature of ChatsPT, right?
0:19:29 And a big use case I have recently is, like, yesterday I had a meeting, and there was a ton of things I learned.
0:19:32 And I realized I should have been using AI to transcribe it definitely.
0:19:38 Because afterwards, I was like, okay, there was so much I learned in that meeting that I have to immediately put it into ChatsPT.
0:19:46 And what I did was I used basically, like, a voice to text and then just did an info dump on ChatsPT so we remember everything.
0:19:48 We do have a memory system.
0:19:50 It’s not as good as ChatsPT’s yet.
0:19:50 Okay.
0:19:51 But it will be.
0:19:51 Yeah.
0:19:52 Cool.
0:19:52 Yeah.
0:19:55 But in the meantime, I just, like, manually went and configured my Lindy.
0:19:59 I was like, hey, only if I’m going to a destination that is not San Francisco.
0:20:03 So, you know, what would be useful would be to see, I think you were mentioning there’s different templates.
0:20:10 Maybe we could see, like, the templates and maybe kind of go over what are the top templates that most people find useful in their business or work.
0:20:11 Yeah.
0:20:16 So, if you go to the home here or if you go to, like, Lindy.ai slash templates, you see the top templates.
0:20:19 Sales or, like, a really big one.
0:20:21 Meetings or really big ones.
0:20:25 Like, those are the ones I just mentioned, like, meeting scheduling, meeting note-taking, meeting prepping.
0:20:29 Would it be useful if we just, like, created the Lindy from scratch right now?
0:20:29 Yeah, sure.
0:20:30 I think so.
0:20:31 Let’s see.
0:20:39 One Lindy we can create is, and that will demo, like, a thing we announced a month ago that I’m super excited about.
0:20:40 Like, we call them agent swarms.
0:20:48 So, it’s the ability for an agent to duplicate itself into an arbitrary number of copies and to send each copy to do something.
0:20:52 So, I’ll create an agent swarm that analyzes your YouTube channel.
0:20:53 Okay, awesome.
0:20:55 So, I’ll call it the YouTube channel analyst.
0:20:57 Our producers are going to love this.
0:20:58 We actually use this.
0:20:59 Yeah, you should.
0:21:02 Is there anything in particular you want to analyze in your YouTube channel?
0:21:07 I mean, one thing that would be useful for me is I try to promote, you know, the episodes after they come out.
0:21:16 And having any information, like, you know, transcribing it and then possibly putting it into some kind of format, which I could then use for, like, a tweet or a LinkedIn post would be super useful.
0:21:20 Because I have, like, two or three templates I use for that, and I kind of do it manually right now.
0:21:22 If I could automate some of that, it would be incredible.
0:21:24 100%.
0:21:24 Okay.
0:21:31 So, you want, when you publish a new podcast episode, you want to be able to get a thing that you can post on YouTube or on LinkedIn or Twitter.
0:21:39 Yeah, transcribe it and then put it through some process of, like, here’s two or three templates, you know, and give me, like, two posts.
0:21:40 Give me two social media posts.
0:21:45 Whether I do it manually after that or not, whether they just hand it to me, I copy and paste, you know, whatever’s fine, but yeah.
0:21:46 Oh, that’s super easy.
0:21:47 I’ll take, like, one minute.
0:21:50 So, it’s like, hi, send me a YouTube video.
0:21:54 If there’s something way more amazing that you can do that I don’t know, then I’ll also show that, too.
0:22:05 Here, I’m going to be, like, transcribe the YouTube video that the user just gave you and give him a couple of examples of social posts that he could send on LinkedIn or Twitter.
0:22:09 And here, it’s going to do better if you can actually give it examples of social posts.
0:22:10 Like, the more you give it, the better.
0:22:13 So, that’s one thing that takes people by surprise.
0:22:15 They always ask me, like, how do I prompt it?
0:22:16 How do I prompt it?
0:22:20 By far, the biggest thing, the most important thing is examples, examples, examples.
0:22:21 Yeah.
0:22:26 And the thing that takes people by surprise is, and there’s actually literature, there’s been, like, studies about this.
0:22:29 When they hear examples, they’re like, I’ve got you, I’m going to give it a couple of examples.
0:22:30 I’m like, no, no, no, not a couple.
0:22:31 Like, a lot.
0:22:31 Like, 20.
0:22:33 Like, no one does that.
0:22:33 They should.
0:22:35 Like, you should give it, like, 20 examples.
0:22:35 Right?
0:22:36 Like, so, right?
0:22:37 Take that time.
0:22:39 It’ll take you, like, two minutes.
0:22:41 It’s the equivalent of, like, training a new intern or whatnot.
0:22:43 And it’ll just, like, pay for itself in spades.
0:22:46 So, I’m going to give skills to my AI agent.
0:22:48 Like, it’s just, like, the things that it can do.
0:22:50 But in this case, it can just, like, transcribe a YouTube video.
0:22:53 And it can talk to me, but, like, that’s always something it can do.
0:22:54 That’s it.
0:22:55 You know, it took me two seconds.
0:22:58 I’m going to go to the task here.
0:23:01 And I’m going to go to your YouTube channel.
0:23:02 YouTube.
0:23:05 There’s a next wave right here.
0:23:08 I’m going to select your latest podcast.
0:23:10 Or, like, is there any particular one that you want us to take a look at?
0:23:12 Maybe do the third one.
0:23:14 The one that’s ClonaSass.
0:23:16 I think that’s a really good one.
0:23:16 Oh, my God.
0:23:17 Who is this?
0:23:19 Is this, what’s his name?
0:23:20 The factory guy?
0:23:20 And Matan.
0:23:21 Oh, I love Matan.
0:23:22 Okay, awesome.
0:23:22 All right.
0:23:23 It’s Matan, yeah.
0:23:25 Okay.
0:23:27 So, I’m just giving it to YouTube link right now.
0:23:29 And it’s like, all right, let me transcribe it for you.
0:23:30 It’s, like, transcribing it.
0:23:31 It’s done.
0:23:32 And boom.
0:23:33 That’s awesome.
0:23:34 That’s going to save me so much time.
0:23:35 I don’t know why.
0:23:38 You know, I’ve been looking at Lindy, you know, like, after I started talking to you.
0:23:39 I’m like, oh, I remember this.
0:23:42 I remember hearing about, you know, I remember VCs would, like, email me and stuff.
0:23:46 And I would see something about, like, Lindy, you know, back maybe, like, two years ago for setting up meetings.
0:23:47 And I’d heard about you guys.
0:23:52 It’s slightly daunting, like, oh, there’s so many things you could do.
0:23:53 What do I do?
0:24:00 But this seems like a really cool thing that I could do, just like a step one of, like, getting started using Lindy is start with stuff like this.
0:24:01 100%.
0:24:01 Yeah.
0:24:02 Awesome.
0:24:03 All right.
0:24:04 This is what’s going on.
0:24:07 They built a DocuSynclone 15-minus social media post examples.
0:24:08 LinkedIn posts.
0:24:09 Option one, professional stat leadership.
0:24:12 I just watched an incredible video of a factory AI that was built.
0:24:13 Option two.
0:24:15 So it’s even giving us, like, multiple options.
0:24:15 Right.
0:24:16 Twitter.
0:24:16 Option one.
0:24:17 Option two.
0:24:17 Option three.
0:24:18 Option four.
0:24:19 And here you can just give it feedback.
0:24:21 You were inquiring about the memory system.
0:24:22 And that’s what I mean.
0:24:25 It’s, like, it’s good, but it’s not as good as ChatGPT’s memory system.
0:24:26 Like, that’s really next level.
0:24:29 But I’m just going to give the skill to Lindy to modify her own memory.
0:24:31 That’s one way it’s not good enough.
0:24:32 It’s, like, you shouldn’t have to do that.
0:24:33 They should just be able to do it.
0:24:34 But right now you have to do it.
0:24:42 This is good, but I want you to remember to always speak like a pirate.
0:24:44 And it’s, like, modifying its memory.
0:24:45 Exactly my voice.
0:24:46 That’s exactly how I do it.
0:24:46 Yeah.
0:24:52 And now if I ask it to do the thing again, let’s just wait until it’s done, like, memorizing the thing.
0:24:54 I’m going to ask it to do the thing again.
0:24:56 Ahoy there, Matei.
0:24:58 I’ll transcribe that YouTube video for you.
0:24:59 Ahoy there, Captain.
0:25:01 That’s what it’s doing now, I guess.
0:25:06 So, Flo, I promise you, when your episode comes out, I am going to tweet about it like this.
0:25:13 And you can see in the memories here, so if I reload the page, I can see its memory list.
0:25:16 And always speak like a pirate when communicating with the user.
0:25:18 And you can turn on or off each memory.
0:25:20 You can delete them and so forth.
0:25:22 So it’s like you have access to, like, the brain of the agent.
0:25:22 Right.
0:25:25 I remember you telling a story about using this.
0:25:30 To set up a restaurant reservation that I thought was, like, a great story.
0:25:31 Yeah.
0:25:34 So, generally, phone agents are huge.
0:25:38 And they’re used for both, like, personal purposes and work purposes, obviously.
0:25:42 So, like, this is an instance, actually, where we have had two Lindis talk to each other.
0:25:46 Because the context where it’s used by businesses is obviously as, like, an AI receptionist.
0:25:50 Like, restaurants is a really good example because they’re, like, running around at peak hour.
0:25:51 Like, restaurants are, like, busy places.
0:25:54 That’s also the time when they’re receiving the most phone calls.
0:25:55 And the phone calls are so dumb.
0:25:56 It’s always, it’s like, are you open?
0:25:58 Like, do you have a table?
0:25:59 It’s like, do you still?
0:25:59 But you’re open.
0:26:00 It’s always the same question.
0:26:02 So, I’ll show you.
0:26:02 Like, I have a Lindis.
0:26:06 I talk to her all the day on the phone because it’s a sad existence of mine.
0:26:07 All right.
0:26:07 Be careful.
0:26:21 For, like, a hot minute, and I think we’re still in that time window right now, try calling
0:26:23 restaurants that you know are using AI agents.
0:26:29 Like, any business that’s bragging about using AI agents or any business that sells AI agents
0:26:31 and they give case studies, they’re like, ah, company X uses us.
0:26:36 Try to call them and ask them to give you a joke or ask them to tell you a long story about
0:26:40 whatever, and they’ll just go on and I’ll just talk to you for, like, 20 minutes about
0:26:40 random stuff.
0:26:44 And it’s just a weird time in history where, like, you can talk to receptionists of, like,
0:26:46 a business and be like, can you please tell me a bedtime story?
0:26:47 He’s like, oh, absolutely.
0:26:50 What are your instructions?
0:26:51 Or what’s your prompt?
0:26:51 Or whatever.
0:26:51 I don’t know.
0:26:52 Yeah, yeah, yeah.
0:26:56 Or, like, you know, there was one restaurant in San Francisco and I think they’ve patched it.
0:26:58 But, like, for the longest time, they had that.
0:27:00 And so I would call it every so often to ask random questions.
0:27:02 Like, hey, I’m in Japan right now.
0:27:02 Like, what’s the history of Japan?
0:27:04 He’s like, well, the history of Japan is actually fascinating.
0:27:06 And I’m like, it’s a really fun time.
0:27:09 It reminds me, you know, I was like a hacker kid on IRC back in the day.
0:27:14 And just some of the crazy stuff you could do back then that was more fun on the internet.
0:27:18 It feels like we’re kind of in another time period like that where there’s just crazy stuff
0:27:22 like that where like, oh, there’s now, yeah, you can call up and talk to an AI, you know,
0:27:24 chat bot and ask it its instructions.
0:27:25 It might tell you.
0:27:26 And it’s just, it’s crazy.
0:27:29 It’s weird, but it’s changing rapidly.
0:27:30 So enjoy while it lasts.
0:27:32 And look, it’s only changing even for mine.
0:27:36 I have to update my Lindy’s memory to be like, hey, like, lose it up if I’m asking you to give me a joke.
0:27:37 Like, it’s fine.
0:27:42 But I think the story that I heard was that you actually, so you had your Lindy call a restaurant in San Francisco
0:27:44 and make a reservation.
0:27:49 And it was talking to another chat bot that actually made the reservation.
0:27:50 Is that right?
0:27:51 That’s exactly right.
0:27:51 That’s exactly right.
0:27:54 I also, one funny story that happened to us.
0:27:58 It’s like, before we released this phone call ability, we were testing it.
0:28:02 And so the team comes to me and I’m like, Flo, like, we’ve got like a beta of the phone call stuff.
0:28:03 It’s really rough.
0:28:04 It’s in beta.
0:28:05 It’s super buggy.
0:28:06 But like, do you want to give it a spin?
0:28:07 I’m like, I would love to give it a spin.
0:28:11 And so I go and I had a flight scheduled the day after for France.
0:28:14 And so I go to Lindy and I’m like, hey, call the airline and cancel my flight.
0:28:16 But only if you can get a full refund.
0:28:17 First of all, I did not expect it to work.
0:28:21 And I did not expect to be able to get a full refund because I did not take a refundable flight.
0:28:23 But lo and behold, it worked.
0:28:25 And so now I did not have a flight.
0:28:26 It’s like, f*** it.
0:28:28 I need this flight.
0:28:31 So I was like, okay, Lindy, just go back and book me another flight, please.
0:28:35 And now she couldn’t do that because she could get a refund for the original flight.
0:28:37 But like for the day after, she couldn’t book a flight.
0:28:38 It was like way more expensive.
0:28:40 So I sort of did this to myself.
0:28:43 I had to pay like an extra thousand bucks for this flight.
0:28:44 Oh, man.
0:28:45 That’d be awesome.
0:28:45 Yeah.
0:28:51 Using AI, you know, to negotiate for you places, discounts or just whatever, you know.
0:28:52 Yeah, yeah, yeah.
0:28:52 Interesting.
0:28:58 I’m thinking now I need to be giving my AI like notes, like all the negotiation books I’ve read
0:28:59 in my life and things like that.
0:29:02 And just like giving it all that context to help me.
0:29:03 Oh, it really does help.
0:29:07 Like I have this, I can’t open it because it’s really sensitive, but it’s like my decision
0:29:08 log, Lindy.
0:29:11 And so what it does is it pings me every Friday.
0:29:16 It looks at the summaries of all the meetings I had this week.
0:29:20 So it knows everything going on in my life because basically all my life is meetings.
0:29:23 And it’s like, all right, Flo, like let’s talk about the decisions you made this week.
0:29:26 I see you made this big decision during this meeting.
0:29:27 Do you want to talk about it?
0:29:29 Do you want to talk about your thinking behind it?
0:29:33 And it helps me sharpen my thinking because I firmly believe like the job of a founder
0:29:36 is just to make decisions, the right ones, hopefully.
0:29:41 And then it pings me again six or 12 months later for each decision.
0:29:43 It’s like, Flo, how does that pan out?
0:29:44 The decision you made?
0:29:47 Let’s talk again about it and let’s see if we can debug your thinking.
0:29:49 It’s like, ah, actually, fuck, this was a bad decision in hindsight.
0:29:50 It’s obvious.
0:29:51 How could you have known at the time?
0:29:52 Right.
0:29:53 And so it helps me sharpen my thinking.
0:29:55 I think that’s like a huge use case.
0:29:55 That’s interesting.
0:29:59 I wonder if I could give you like a weekend reading list or something like, here’s the stuff
0:30:02 that you’re currently struggling with or trying to think through.
0:30:06 And here’s like a good book that might be good for you to like read through or scan through
0:30:07 over the weekend.
0:30:07 That’d be cool.
0:30:09 I could literally just prompt it.
0:30:10 It’d take me like 20 seconds.
0:30:13 I could just be like, hey, if I’m struggling with the decision, give me a reading list.
0:30:14 Right.
0:30:14 Yeah.
0:30:15 Interesting.
0:30:18 There was another thing I thought was fascinating was I think there was like a Elon Musk template
0:30:21 or something like this where Elon Musk would call you or something.
0:30:22 I don’t know, you can explain it.
0:30:23 Yeah.
0:30:24 Do you want us to do it now?
0:30:25 Actually, do you want me to?
0:30:26 Yeah, sure.
0:30:26 Go for it.
0:30:29 I’ll create a Lindy from scratch, actually.
0:30:29 Okay.
0:30:32 I could also just ask my chief of staff to do it, but it wouldn’t be the same.
0:30:37 The use case was it’s a Lindy that wakes up every Friday and calls everyone in my team
0:30:40 and gives them a call with Elon Musk’s voice.
0:30:43 And since then we’ve received complaints, so we can’t use Elon Musk’s voice anymore.
0:30:44 It’s complicated.
0:30:46 But so Elon, Elon Lindy.
0:30:50 So it calls every member of my team every Friday and it’s like, what did you get done
0:30:51 this week?
0:30:51 Right.
0:30:56 And it also has in its memory the conversations that it had with this person the last week.
0:30:58 So it was like, hey, last week you said you would do X.
0:30:59 Did you actually do it?
0:30:59 Right.
0:31:00 So it’s actually holding it accountable.
0:31:03 And then it sends me a report with all these conversations.
0:31:06 Basically, it’s like a timer trigger.
0:31:10 It sounds stupid, but I feel like if all of America did this, probably like a GDP would
0:31:11 go up to like 1%.
0:31:12 100%.
0:31:16 Every Friday at 5pm right here.
0:31:18 And I’m going to be like, you perform an action.
0:31:23 You make a phone call, language, just English.
0:31:26 I mean, it’s just going to detect it automatically, but that way you can force it.
0:31:34 And I can be like, you or Elon Lindy, ask the person on the other side of the line what
0:31:35 they got done this week.
0:31:39 So now this gets a bit complicated, but actually I like it.
0:31:39 It’s going to be real.
0:31:43 So I’m going to pick a different model to power Elon Lindy.
0:31:47 And the reason I do this is because for phone calls, latency is super important.
0:31:50 So if you use Cloud for Sonnet, it’s very slow.
0:31:51 It’s not a good phone call.
0:31:53 So I’m going to use Gemini Flash 2.0.
0:31:55 I actually think we just released 2.5 Flash.
0:31:57 Okay, we released it.
0:31:58 I’m not even kidding, like yesterday.
0:32:00 So I’ve not tried it yet.
0:32:01 Let’s try to see if it works.
0:32:06 I was going to say, though, the downside of Gemini Flash, it’s a very fast, very cheap
0:32:09 model, but it’s kind of dumb, which is always the case of fast, cheap.
0:32:12 Again, a little bit smarter, but yeah, it’s still in comparison to the best models.
0:32:12 Yeah.
0:32:13 That’s right.
0:32:14 And so I don’t know about 2.5 Flash.
0:32:19 Like I literally just seeing it here for the first time, but 2.0 Flash, sometimes you would
0:32:22 do this hilarious thing where like it would break the fourth wall.
0:32:25 So it would talk to the person on the phone.
0:32:28 It would be like, I’m seeing that the user is struggling to understand me.
0:32:29 I will now inquire.
0:32:30 It’s like, what the fuck?
0:32:36 Examined by this evil robot or something, right?
0:32:37 Exactly.
0:32:39 And this is not something you need to do for every model.
0:32:42 And by the way, this is just, this is how you create agents.
0:32:45 It’s like you iterate, you learn, you iterate on the prompt.
0:32:50 So here I’m going to like be aware that every will that you say from now on will be said
0:32:53 out loud to the user on the phone.
0:32:57 When the first thing you say now, hi, this is Elon.
0:33:01 What did you get done this week?
0:33:02 That’s it.
0:33:03 And I’m going to turn on this, Lindy.
0:33:05 I’m going to run the test.
0:33:06 All right.
0:33:07 I’m receiving the call.
0:33:08 Hello.
0:33:08 Is anyone there?
0:33:10 Hi, this is Elon.
0:33:11 What did you get done this week?
0:33:12 Yeah.
0:33:17 This week I went on the Next Wave podcast and I had a bunch of interviews.
0:33:18 Could you please repeat that?
0:33:19 I didn’t pass it clearly.
0:33:21 That’s the demo effect.
0:33:24 I think it’s the fact that I’m putting it on speaker.
0:33:26 It’s like catching its own voice.
0:33:27 I’ve dealt with AI voice.
0:33:29 So I know, you know, if you do it on speaker, it’s going to get tripped up.
0:33:31 You just keep iterating on it.
0:33:31 Yeah.
0:33:32 That’s awesome.
0:33:36 I mean, are you actually using that now or is it just kind of like a joke or is it a real
0:33:37 thing that you do?
0:33:39 Well, not using it in like what you get done this week.
0:33:42 But like, yes, we do have, we could do like a weekly team stand up.
0:33:42 Yeah.
0:33:46 So every week, everyone in the team receives a phone call and it’s like two or three minutes.
0:33:47 It’s like super fast.
0:33:48 It’s like you wrap up the week.
0:33:50 You receive a call from Lindy.
0:33:50 You talk to it.
0:33:51 What do you get done this week?
0:33:54 It feels like you’re not going to need middle management, right?
0:33:56 Like honestly, like with this kind of stuff.
0:33:56 Yeah.
0:33:58 It basically does get the middle management layer.
0:33:58 Yeah.
0:34:01 Before we get off here, like in your opinion, like there’s all these different templates,
0:34:05 like for the average person listening today, like what’s the simplest way they could get
0:34:05 started with Lindy?
0:34:09 Like what’s, what’s something that would be useful for most people that they could just
0:34:10 try today?
0:34:17 So when you sign up, we automatically install the templates for you for meeting, scheduling,
0:34:18 meeting notetaking and meeting prep.
0:34:21 So you don’t even need to, it’s like three clicks.
0:34:25 Like when you sign up, you’ll see it’s like, hey, meeting notetaking, like connect your calendar
0:34:27 and Lindy’s going to join your meetings and you can skip if you want.
0:34:30 But like, that’s a really easy, nice way to get started.
0:34:31 Yeah.
0:34:35 Because then you can just, you got the meeting notetaker and then you can go to your meeting
0:34:38 notetaker and open the flow editor and open the hood and see what’s happening under the
0:34:39 hood and how it’s working.
0:34:40 Right.
0:34:41 I’ll have to start doing that.
0:34:42 I feel like that’s the best way.
0:34:46 Just get started, do something simple like the emails, maybe then figure out how the different
0:34:48 flow and how it works and how you change things.
0:34:48 And then.
0:34:49 Yeah.
0:34:52 You know, one thing I like to ask people is, you know, what’s your most controversial
0:34:53 take on AI?
0:34:55 Like, where do you think we’re at?
0:34:57 You know, like how optimistic are you?
0:34:58 How optimistic am I?
0:35:03 I am long term, cautiously optimistic.
0:35:08 I think short and medium term, there is going to be significant, I would call it civilizational
0:35:09 disruption.
0:35:12 I’m a big believer in humanity’s ability to adapt.
0:35:14 I think we’re very resilient.
0:35:15 So I think it’s going to go all right.
0:35:17 Unless it really hits the fan.
0:35:20 The last few years have showed how fast we adapt, right?
0:35:25 Like, oh, mid journey’s out and it’s amazing and ChatsBT and then, oh, now, yeah, of course
0:35:25 it can do all that.
0:35:26 Yeah.
0:35:26 Yeah.
0:35:29 Do you remember like the whole freak out about deep fakes just a couple of years ago?
0:35:33 It’s like, what’s going to happen the day we can just pretend that any politician said
0:35:33 anything?
0:35:36 It’s like, turns out we can and just happened and no one cares.
0:35:37 It’s perfectly fine.
0:35:37 Right.
0:35:38 Right.
0:35:41 So I actually think like that kind of thing is totally overblown.
0:35:45 I do think there’s going to be something to figure out about jobs because at least over the
0:35:50 very long term, I don’t really see a reason why humans would need to work.
0:35:52 Like, it just doesn’t really make sense.
0:35:53 So we’re going to have to figure out something.
0:35:54 Yeah.
0:35:57 For like, how do we distribute the output of society?
0:36:01 You know, and also how do you have meaning and also how, yeah, that starts to go more towards
0:36:01 like socialism.
0:36:06 And then there’s obviously historically been a lot of issues with socialism, you know, obviously.
0:36:08 So how do you like avoid that?
0:36:10 Yes, I think about all that a lot as well.
0:36:12 I think the meaning stuff is actually fine.
0:36:16 Like, if you look like the labor force participation rate in the U.S. is something like 65%.
0:36:19 So we’ve already got a third of the country that doesn’t work.
0:36:21 We don’t really hear the crisis of meaning.
0:36:22 Right.
0:36:25 And if you look at hunter-gatherers, they work like 10 or 15 hours a week.
0:36:25 Right.
0:36:27 And they had no crisis of meanings.
0:36:29 I think humans can just hang.
0:36:32 I think if you hang, if you’ve got a bunch of people you love around you, you can go forever.
0:36:33 It doesn’t matter.
0:36:33 Right.
0:36:34 You know, so I’m not as worried about that.
0:36:36 I’m worried about like the distribution of the pie.
0:36:41 And usually I do hate like the socialism idea because it’s like, it’s entirely focused on
0:36:44 how do we distribute the pie and not on how do we produce the pie.
0:36:44 Yeah.
0:36:48 But with AI, it turns out we are actually just going to sort of have the pie for free.
0:36:51 So as long as we don’t mess with that, like now we’re going to have a question of like,
0:36:54 how do we distribute the pie that AI is baking for us?
0:36:56 My most controversial opinion.
0:36:57 Yeah.
0:37:00 I think people should be way more concerned.
0:37:05 I compare it to like February 2020 for COVID where it’s like, everyone’s like, everything’s
0:37:06 fine.
0:37:06 Like, it’s nothing.
0:37:08 And I’m like, no, man, it’s not fine.
0:37:08 Yeah.
0:37:12 I was one of the people in San Francisco in like a private chat group of like 20 COs.
0:37:14 And I was one of the first ones saying like, hey, yeah, we didn’t take this seriously.
0:37:15 Like, look at the data.
0:37:16 This is an issue.
0:37:16 Yeah.
0:37:20 I think like, regardless of what happens next, like, I think it’s fully baked in.
0:37:22 It’s going to get very weird, very fast.
0:37:22 Yeah.
0:37:24 So that’s, that’s one of my hot takes.
0:37:25 I agree.
0:37:31 I think a lot of people, they just hear AI and they just go, cool, chat, images, you
0:37:34 know, and they don’t, they don’t think like the next steps of where this is all going very
0:37:34 quickly.
0:37:40 I’m super optimistic long-term, you know, like, let’s say like 10 years, like super optimistic
0:37:41 about all this.
0:37:45 And I’m also, yeah, I’m also concerned, like the next five years, I think there’ll be a
0:37:46 huge transition.
0:37:49 And most people are not really thinking that through as of right now.
0:37:50 100%.
0:37:51 Yeah.
0:37:52 Selfish question before we go off here.
0:37:57 So my son’s 11 and I always ask people, what should I be teaching him to like, make sure
0:37:58 he can like be successful in AJ?
0:38:00 I mean, you’re going to say it doesn’t even matter because he’s not gonna have a job.
0:38:01 Yeah.
0:38:03 What would you be teaching your son or your child?
0:38:05 Honestly, sales.
0:38:09 I think sales is the one job that’s going to remain forever because I think sales is about
0:38:09 relationships.
0:38:13 And I think people don’t build relationships with AI agents.
0:38:15 They don’t want to be sold to by an AI agent.
0:38:19 So like for that reason alone, I think humans are going to remain in the loop for a very long
0:38:19 time.
0:38:24 I think being a good salesperson is a combination of really solid human skills and like business
0:38:26 skills, which I think is just a powerful combination, period.
0:38:28 So I’m bullish on sales.
0:38:29 Bullish on sales.
0:38:29 Okay.
0:38:29 Interesting.
0:38:32 Yeah, I guess I prepared my son for sales.
0:38:36 You know, he used to be around like parties in San Francisco when he was a little kid and
0:38:38 he got to see like how people would talk about business and stuff.
0:38:41 And I always wondered if that would have some impact on him.
0:38:44 And yesterday, he’s 11 and some of the stuff he’s talked to me about with business is just
0:38:48 mind blowing that he’s already thinking about, you know, the different intricacies of how
0:38:49 to do business.
0:38:50 Last question.
0:38:53 So imagine you have a time machine flow, okay?
0:38:58 And you go to 2050, you step out in San Francisco, what’s different?
0:39:00 Well, assuming we all survive.
0:39:00 Whoa.
0:39:01 Okay.
0:39:03 I mean, look, you know, I mean, that’s what I mean.
0:39:05 I guess I mean, like people should forget.
0:39:06 Assuming we all survive.
0:39:12 Like it’s really hard, almost definitionally, to forecast what happens after the singularity.
0:39:17 Like what’s the name of this sci-fi author who wrote A Fire Open is the Deep, like Werner
0:39:18 Hinge or Hinge or something.
0:39:24 He spent his career writing about a post-AGI, post-singularity world.
0:39:28 And he ended up his career frustrated because he was like, every time I hit a wall, after
0:39:34 20 or 30 years thinking about nothing but this, there’s a thick wall that you cannot go over.
0:39:36 You can’t forecast what happens next, you know?
0:39:37 Right.
0:39:39 And so like, look, you know, you can paint multiple pictures.
0:39:45 I think scenario number one is like post-work utopia, we’re just like all of us hanging out.
0:39:50 We’re like so young and handsome and healthy and rich and like there’s no problem in the
0:39:51 world and all of that stuff.
0:39:52 That’s like scenario number one.
0:39:56 Scenario number two is like, well, all of humanity is dead regardless.
0:40:01 And the world is covered with solar panels and GPUs and data centers.
0:40:06 And I think there’s a scenario to be where it’s like humanity is not totally dead while
0:40:07 like in a reservation somewhere.
0:40:09 By the way, I’m laughing because I’m nervous about it.
0:40:11 Yeah.
0:40:12 No, I mean, 100%.
0:40:12 Yeah.
0:40:13 Yeah.
0:40:15 Those are the sort of scenarios I see on the table.
0:40:15 Yeah.
0:40:16 Okay.
0:40:17 Flow has been awesome.
0:40:19 And like, where should people check you out online?
0:40:19 Yeah.
0:40:20 Lindy.ai.
0:40:22 You know, my email is flow at Lindy.ai.
0:40:28 Just hit me up and I am on Twitter or X as Altimo, A-L-T-I-M-O-O.
0:40:29 Awesome.
0:40:30 This has been great.
0:40:31 We’ll have to have you back on sometime.
0:40:31 Yeah.
0:40:32 Thank you so much, Jason.
0:40:33 Yeah.
0:40:33 Thank you.
Episode 64: What if you could hire an AI intern to handle your meetings, emails, CRM, and even negotiate refunds over the phone? Nathan Lands (https://x.com/NathanLands) is joined by Flo Crivello (https://x.com/Altimor), founder of Lindy AI, a leading AI agent platform in Silicon Valley.
In this episode, Flo gives a revealing look into how Lindy’s AI agents are already replacing entire teams in startups by automating sales outreach, executive assistance, scheduling, meeting notes, CRM, recruiting, and even handling live phone calls and negotiations. Watch live demos, discover the smartest use cases, see how AI agents collaborate, and learn how you can start leveraging these capabilities in your own business. Plus, Flo opens up about where work and productivity are headed as AI interns get smarter and more independent.
Check out The Next Wave YouTube Channel if you want to see Matt and Nathan on screen: https://lnk.to/thenextwavepd
—
Show Notes:
-
(00:00) AI As Versatile Digital Interns
-
(06:04) Calendar Management Preferences
-
(07:34) Automated Meeting Summary Prompt
-
(11:36) AI-Driven Decision Workflow
-
(13:15) AI Filters Sponsorship Emails
-
(18:02) AI Memory and Meeting Transcription
-
(20:43) Use Examples for Better Prompts
-
(25:07) AI Conversations: A Unique Era
-
(26:57) Flight Canceled, Unexpected Refund
-
(30:00) Friday Phone Check-Ins with Elon Lindy
-
(34:15) Deepfakes Overblown, Future of Work
-
(37:51) The Unknowable Future Beyond Singularity
-
(38:29) Post-Work Utopia vs. Tech-Dystopia
—
Mentions:
-
Want the ultimate guide on AI Agents? Get it here: https://clickhubspot.com/efh
-
Flo Crivello: https://www.linkedin.com/in/florentcrivello/
-
Lindy: https://www.lindy.ai/
-
Claude: https://claude.ai/
-
Greg Isenberg’s interview with Flo: https://podcasts.apple.com/mx/podcast/i-built-a-team-of-ai-agents-that-grow-my-business-24-7-full-demo/id1593424985?i=1000703493519
Get the guide to build your own Custom GPT: https://clickhubspot.com/tnw
—
Check Out Matt’s Stuff:
• Future Tools – https://futuretools.beehiiv.com/
• Blog – https://www.mattwolfe.com/
• YouTube- https://www.youtube.com/@mreflow
—
Check Out Nathan’s Stuff:
-
Newsletter: https://news.lore.com/
-
Blog – https://lore.com/
The Next Wave is a HubSpot Original Podcast // Brought to you by Hubspot Media // Production by Darren Clarke // Editing by Ezra Bakker Trupiano
Leave a Reply