AI Agencies Just Got Simple Enough for Anyone to Start⏐Ep. #223

This is a service you could offer for a monthly fee. Just using Lindy on the backend.

It's so easy to create AI agents, it's almost addictive. It's literally just you type words and just does it for you. You can have the entire flow completely end to end managed by AI agents.

There's a dozen Silicon Valley companies right now, raising tens of millions that are basically just that agent.

People underestimate how early it is. The capabilities that we're talking about today are literally 30 days old. There is an immense opportunity for these AI agencies.

The people that I've spoken to that want to start an AI automation business, they're getting tripped up because they're not technical. I'm not technical, but I like this, removes that friction.

There is no excuse anymore not to use AI agents.

The sooner you act, the more you'll be ahead of your peers. I've said it before and I will say it again, that one of the biggest opportunities I see right now is that of starting an AI automation agency where you insert yourself into small to medium sized businesses and use very cheap, very easy to use tools to automate certain processes that these businesses use. And you charge between $500 and $5,000 a month or you charge 3 to $5,000 one time and 2 to 700 dol dollars a month ongoing. So one month ago Lindy, which is an AI automation platform, came out with an update that enables anyone to create AI agents with one prompt. And this is game changing technology. There is a massive opportunity to be the expert in the room for these business owners that know they need to be using AI but have no idea where to start. So without further ado, please enjoy. We use a couple acronyms in here that some people might not understand. The first one is n 8 n n and then the number 8 and then n, which is a somewhat technical AI automation platform that you really need some base level of technical knowledge to use. So we contrast that with Lindy and then the Other acronym is STR, sales development representative, aka a salesman.

We just released Lindy 3.0. That's like the biggest update in our history as a company. And the three big updates are one, like you can now collaborate with your team on Lindy, like invite your team members and share your agents and so forth. But most importantly, the two really big things that we're really excited about, the Agent Builder. So like now you can create an agent from just a prompt. You literally just like type words on your keyboard and you're like, this is what I want to automate. I want to automate my customer support, my content generation for marketing, my like, lead qualification and whatnot. And just literally from just a prompt like create the agent for you. And computer use, which is the idea that your agents can now use their own computer in the cloud. Like, they have like a, if you will, like a virtual mouse and a virtual keyboard that they can use. So that means that agents are no longer limited by the MCPS or like the integrations that they have access to. Like, as long as this is a tool that is accessible from a web browser, that is something that your agent is going to be able to use, even if it's like an internal tool of yours. So we have an entire teams right now using Lindy, for example, to automate their customer support and have internal tools to issue a refund. And until now these guys were blocked. They were like, ah, we really want to use AI agents, but we don't have an API for these internal tools and we don't have the bandwidth to build them. And now they're actually deploying these AI agents really quickly because these AI agents can just log in and use their internal tools using just like a mice and keyboard. Those are the two very, very big things that have changed.

Okay, that's incredible. So basically, with a Vibe coding tool, with one prompt, you can create a web app or a website. Cool. With this, with one prompt, you can create an agent that will respond to all of your customer service inquiries. Now, it can't be too broad, right? It needs to learn and know how to respond to certain inquiries. But if my prompt were to say, every time I get an email to supportrismail.com that says, I want a refund, please issue them a refund. Right. Would that be a use case? Like, please go into stripe, issue them a refund.

That's exactly right. Yeah. So this is an example. I use Lindy to manage my personal CRM. So it's more like a work personal example. I meet people every day and I want to keep track of them. And so I want to be able to go on Lindy and be like, hey, I met this guy, he's an investor, for example. And then I want to be able to go back to Lindy at any point and I want to be able to ask like, who are the investors that I know? So I just sent it a prompt like, help me manage my personal CRM. And then it clarifies the plan that it came up with and it's like, okay, so if you're asking me to Add a contact. I'm going to add a row to Google Sheets, and if you ask me to retrieve a contact, I'm going to get the data from the Google Sheets and I'm going to give you the response like, is this what you want me to do? So it clarifies, it asks you disambiguity questions, and then I say yes, and it creates the agent. And in a matter of seconds, I have this agent here on the right. If I want to add a contact, this is this. And if I want to retrieve a contact, this is that. It's actually funny. Right before this, I was on another podcast with Eric Hsu, I don't know if you know him. And he was like, oh, hit me up next time you're in la. I'm based in San Francisco. He's based in la. And I just went to my Lindy and I was like, hey, rxu, AI influencer. Hit up when he's in la. So there is another prompt that I sent to this Lindy after this video where I was like, hey. Also detect when I'm flying somewhere by intercepting the flight confirmation email from my inbox. And when I fly somewhere, tell me who I need to meet up there. So you see here, it set up this workflow just like, okay, if you receive an email that's. Oh, let me right here. If you receive an email that's about a flight, then I'm going to get your spreadsheet and I'm going to send you an email with who you should meet there. Oh, man.

So you use Google Sheets as your personal CRM.

Well, I think that's the beauty of AI agents is like, yeah, now you just. I just need like a dumb data source and all of the intelligence on top of that and all of the interface on top of that is just agents.

Yeah, no, I. I'm asking because I love that, because I use everything in Google Sheets and I try to eliminate all these other random softwares. But how could we create that with a prompt like specifically the agent 2. Anytime you see a flight, log it. And then anytime I ask you who's in this area, you'll automatically know so I can know who to visit. How could we create that one?

It's literally just a prompt. Like when you log into Lindy, there is this text box here. You can either type your prompt in here or you can press the mic button to dictate it. We actually recommend dictating it because you can give a lot more details when you dictate. And you would be surprised. You can ramble, you can say anything you want, and agents are actually really good at making sense of what you said. And so you just do that, you prompt it and then like a minute or two later you get an agent.

Before Lindy 3.0, if I wanted to do the same thing, I kind of had to know these tools and I kind of had to know how to drag and drop and tie it together and get the API key. And now you don't have to. You just removed all the friction.

That's exactly right. There is no excuse anymore not to use AI agents. Like we've really gone through that journey of like, at first you had to be an engineer. Like if you're an engineer, like, it's a great tool to use. Then we built this no code platform and so you don't have to be an engineer anymore, but you still have to learn the tool. And ain't nobody got time for that. Right. Like you just want to do your job. And so now it's literally just you type words and just does it for you. And if you want to iterate on it, you'd send it another prompt and you're like, ah, like this is not exactly how I want it and it's just going to get to do it better.

Yeah. How much time do you spend looking at what people are doing with Lindy 3.0, seeing how they're using it? Is that a big part of your day or does your team manage that?

Well? I mean, I talk to users all the time and I look at data as well, so I think it's important to have that like forest view and that like tree view. We also look at like the breakdown of like the kind of tasks that people ask. But yeah, I mean, I keep my ear very close to the ground for sure.

What kind of patterns are you seeing and is anything surprising you?

Last night I was looking at a Lindy that a customer sent me and he was like, hey, he was actually proud of it. He sent it in our Lindy community. I recommend everyone checks it out. We have a really vibrant Slack community with thousands of members and he DM'd me. He's like, ah, check out this Lindy. I'm really proud of it. And I was like, okay. I looked into it and I wish I could share my screen. I don't want to violate this person's privacy. This is by far the most insane Lindi I have ever seen. It's like 400 nodes. Those we call nodes. Like 400 nodes.

Wow.

Yeah, you just keep and keep and keep going. And the guy just locked in and he built 80% of it using the prompt to agent that I mentioned and 20% of it by tweaking it across manually. And he basically built. It's sort of like an sdr. It's an indie that fully autonomously not only finds leads online, enriches them, qualifies them, decides who to actually engage with, researches them, sends them a personalized email and then engages in an insane back and forth with them if people say they're not interested, it asks them why. It handles objections. It learns from the objections, it ab tests the emails and the subject lines. It's like, oh my. And then it's like, yeah. And now I can just run it. And it's like a salesforce for me. I have been surprised at how people get sucked in and they get into it. Once you make it so easy to create AI agents, it's almost a addictive.

Like there are, there's a dozen Silicon Valley companies right now raising tens of millions that are basically just that agent aisdr. Yeah, like that's all the rage right now. And for those listening, sales development agent, like just AI sales through email in this case, right?

Yeah. Like there are a lot of startups that do AI SDRs and like, you know, we have this meme internally, it's like, hashtag should be a Lindy. So we see a lot of startups and we're like, should be a Lindy.

That's exactly what I'm talking about.

Yeah, no, exactly. It's kind of like there is an app for that. It's like should be a Lindy. I think the advantage of using a platform like Lindi, and we hear that all day from our customers, is you have a single pane of glass where you can manage all of your agents and then also a single place where your agents can collaborate. I think you really unlock the magic. Once you have your agents work together, which you can do, then you can almost modularize your agents. You can create agents that are specialized in certain tasks and you can get them to talk to each other and collaborate with each other. And in the end it's going to be more powerful, it's going to be cheaper and it's going to be easier to manage for you to have like one single platform where you can do all of that and have like 20 different logins and platforms that you have to learn and subscriptions to manage for each of your use cases.

I'm thinking of like what this use Case could be for the AI sdr. Specifically, if you are an employee or a business owner, let's say you're an employee that's just sells Salesforce software, right? And I go permissionlessly as just an enterprising entrepreneur on behalf of you, an employee selling Salesforce, making good money, presumably. And I go find a bunch of potential clients, I go create an AI sales development agent in Lindy and I go start emailing. Just cold emailing. Not just cold emailing, but having these automated conversations back and forth with people, right? And then a small percentage of them come down the funnel and they're like, let's hop on a call, okay? So you get it all the way to that point and then you reach out to the Salesforce rep and say, hey, I have a really hot lead for you. It's this guy that works in this company, yada, yada yada. I'm guessing it's going to be worth 4,000 mrr to your company, which is worth X, Y or Z, you know, commissions to you. How much would you pay me for this lead? You're not just like selling a dream, you already have it, right? And now he's interested. Now you have his attention. And your chance of getting a reply to that cold email versus hey, I help Salesforce reps get more leads. Can we help on a call? Like, your chance of getting a reply to that is like half a percent. Your chance of getting a reply to, you know, here's a lead, here's the whole conversation, here's his name, phone, email, take it, steal it, it's all yours. What's your chance of getting a reply to that conversation?

100%. We have a recruiting firm that we're working with that's leveraging agents for a use case, actually very similar to the one you just laid out. Like, what they do is that they do send these outreach to people, but what they do is that they look at the list of job descriptions that the company has on its career page and so they're like, oh, you're looking for software engineers who know React and Typescript and yada yada yada, and have minimum X years of experience and all based in San Francisco, California. So they collect all of these requirements, but they have the to clarify, this is an AI agent that like they've set up. So they've set up this agent that does that. Like it collects a bunch of companies that are hiring, it looks at their open positions, it collects the requirements and then it goes on LinkedIn and finds people that satisfy These requirements. And then it sends in the email a list. It's like, hey, we saw you were hiring, including software engineers. This is an example of like three leads that we found and we have their email address. We can reach out to them. Three leads that we found that satisfy your requirements. We've got a thousand more like those. If you say yes, I can send the contract and we can be reaching out to them today, like in an hour. And they'll ripping it. Like, these guys are just like crushing it. They're getting so much response from this email because to your point, it's like, it's so much more engaging and you're bringing so much more value to the table through these emails and through just like a generic like, hey, do you want to work with us for recruiting email?

Yeah. Because sales like this brings sales from here's what I can do for you to here's what I just did for you.

Will you customer? I love that. Exactly. Yeah.

Oh man. We might have talked about like a version of this idea on our last episode, but just imagine if I had a Facebook ads agency, right? I charge $2,500 a month to any company in the world to manage their Facebook ads, right? You could with one prompt, right? Might need follow up prompts, but you can get the, you know, the skeleton for this in one prompt. Say, hey, I want you to go to the meta ads library and I want you to cross reference with LinkedIn and I want you to find me email addresses for decision makers at any company that has an active ad in the US on meta ads library. And then I want you to send them an email, a customized email that says, what are you doing? You only have four creatives on your meta ads right now. Did you know that, you know Companies that have 40 creatives or more have a 30% drop in return on ad spend or increase on return on ad spend. I can do that for you. Like, and full transparency. I use AI for my creatives, so that's not a bad thing. That means you can get 10x the creatives for 30% less cost. When can we chat?

Oh, you could go further than that. Like you could have the agent like search. I mean, this is great. I love that, right? I'm just like reefing here. Like you can, you can have the agent also research the, the competitors of this company research the Facebook ads that these competitors have running and tell the person. And I love the framing that you just put together. It's like it's no longer like, here's what I can do for you. It's like, here's what I did for you. It's like, hey, I researched your competitors. They're crushing it. These are the five best performing ads. Then you add the screenshot in the email. By the way, I've already created these ads for you. This is the creative that they've created. It's like using the same idea, the same concept for the ads, but with your brand and your company and your product. If you just say yes, I'm going to spin them up on your account, right? So yes, because AI agents have lowered the bar so much and because they can make it so easy for you to do this work at a high volume before you've signed the customer, you can deliver the value before you've signed them.

Oh, man, I just like, no offense, I'm having a hard time even listening to you because my mind is just buzzing with other prompt ideas, right? Hey, Lindy, go to Instagram and find me a US based influencer with at least 50,000 subscribers that doesn't use like a link tree or anything, right? And then send them this message. Hey, I noticed you don't use a link tree. Oh. Then you go to Lindy in a different window and say, hey, build me a link tree competitor with your vibe coding tool, right? And it's like, okay, hey, FYI, I have a free link tree competitor. Don't pay link tree 20 bucks a month. Use mine. You have 50,000 subscribers. You should be selling T shirts or newsletters or something in there. Use this, right? Go. And it just, it starts sending messages.

I literally have open right now. Let me share my screen with you. A Lindy that I created to do basically exactly what you just said. So it's like, hey, like this. Lindy is like, hey there. I'm ready to build you like a laser targeted list of influencers. Like, what are you looking for? And I said like, I want 20 influencers on YouTube with 20,000 to 40,000 subscribers in the productivity and AI space. And you can see like, I made a typo here. I said AI space here. But AI agents are awesome. Like, they don't care about typos. And it's going online, it's using a bunch of tools, it's doing a ton of complexity queries. It's doing a lot of research. It's like looping through all the people it found. If you look here, just going to open a new tab. This is how the flow is set up in the background. I didn't have to set this up. This was set up from just a prompt. And you can see here, for each influencer that it found, it's going to scrape their YouTube to find the videos that they have and to understand the influencer better. And it's going to look for their email address, it's going to do all of that stuff. And then it's like, oh, hey, I found these people for you. So this is the focus. Self development, wellness, productivity. This is their email. This is the partnership contact. Look, this guy already has, like, strong brand partnerships with Opus and Heygen.

Oh, my gosh.

No. 100%, like, what the use case is.

You can ask it to do that.

Yeah, absolutely. Yeah.

You asked it to go a step further and find the brand partnerships that they have.

Yeah. In the original prompt, when we said that, I don't know if we set it up exactly this way. I think we just said like, hey, find us details about this person. And, like, you can see he's got like a bunch of scrapers, like, all right, I'm gonna scrape their TikToks, their LinkedIn, their YouTube, their Instagram, their Twitter. And when he does that, it all defines all of this information by default.

Man, rip to our inbox. Like cold emails, they're just getting out of control.

Then you really go one step further and you set up a Lindy to help you manage your inbox. And so we've already had, actually, because that's. It's funny, actually, we have that a lot. We have, like, brands in particular, E Commerce and like CPG brands, like consumer packaged goods brands that are using Linde for this, for, like, influencer outreach to, like, promote their brand. Then we also have influencers, and I wish I could say who, but some. Some extremely large and famous YouTubers are using us. Because these guys, you know, and I'm sure, you know, you know, you're harassed by offers for sponsorships in your inbox every day. And so these guys, they have agencies that actually pay a lot of money, and all of these agencies do all day is they intercept these emails, basically. And then they archive, like, there's a lot of randos and, like, unserious people in there. And then they engage with serious people and serious brands that actually have money and they negotiate a deal. They're like, hey, you know, Bob can work with you. Just a heads up, like, every video is $100,000 or whatever. Not just one, but multiple really, really big and famous, like, YouTube influencers right now using us in this context. The funny thing is that we have actually all the ad Instances of these two Lindy's talking to each other. So like a brand reached out to this influencer using Lindi and this influencer Lindy engaging with that Lindi to negotiate the deal.

How do you know that's happening?

Because we were wondering, well, like this influencer, like, do you think they're going to be in touch? And so we set a reminder to check a month later and we looked into our database and sure enough, like, the two Lindies got in touch with each other.

It's funny, I'm laughing because I, I have a Lindy for that exact same use case because I get these spammy emails, let me sponsor your channel. And I just respond to them with like a number that's way, way too high because usually I just delete them. I love that I'm archive them. So it's just like I'm go, either I ignore it or I'll just give them a stupid number and see what they say.

Yeah, a suggestion. Then you could go again one step further. And like, as you offer a number, you could also research the brand first. Right. And so you can reply with a number but with a way that's like, just like a tiny bit more tailored. You could be like, hey, by the way, our audience is quite aligned with like, who you seem to be selling to.

Yeah, yeah. Do you offer like a white labeling service?

We don't. No. Like all brand is all brands. It's a Lindy. Yeah. Well, it depends. What do you mean by white labeling service?

I'm just picturing like chrisroofingautomations.com right. And I start a business that's basically just Lindy white labeled. And I go into roofing companies and say I can automate any of your processes and save you $30,000 a month, yada, yada, yada. Is that like a growth opportunity for you guys?

100%. So we don't offer like white labeling per se, but we do one, we do let you deploy a chatbot on your website that's like, hey, it's like an AI chatbot. So that's one thing, but also what you just described, like, AI agencies. And I think we talked about that the last time you and I talked, like, AI agencies are huge right now. And look, we do have this new prompt to agent product, but it's still not as good as a human. You're still going to need consultants to really understand how you can deploy this in your business. And it still takes time to build these agents, even if you no longer need the Expertise to do it. It still takes time. I do think there is an immense opportunity for these AI agencies. And we have a couple of AI agencies that we're working with that are extraordinarily successful and growing super fast right now.

We talked about that in June. Is there any update there? Like, how people are using linde to make money and, like, what does their pricing look like, how are they finding customers, et cetera.

So I think that's the beauty of these agencies is, like, each of them is an expert in their own vertical and each of them has their own existing network. So, like, some of them sell, like, sales services, like AISDL services, if you will. Some of them sell to, like, healthcare companies. Like, we have one that, like, interestingly enough, sells to, like, dentists. And they were just really specialized in, like, hey, I know how to automate your back office as a dentist. Apparently, dentists, they're still running, like, extremely antiquated software. And you can automate so much of that using AI agents. So what I recommend to people is the value that you are delivering as an AI agency is that you are bridging your expertise in the technical product. Like Lindy, you know how to set up and monitor these AI agents and your expertise in any given business or any given area. And I think people consistently underestimate the value of their expertise in the latter. They sort of take it for granted. They're like, everyone knows about that. It's just like, no, man, you're actually in a really interesting niche right now. And don't underestimate how few people live in the intersection of these two things. Many people know about AI agents. Many people, some people at least know about AI dentists and the software issues and back office issues that they're having. Few people live in the intersection of those two. And when you do, like, you can. You can deliver a lot of very differentiated value.

Yeah, well, a lot of times people like, like you and I or people that are playing around with these AI tools are just hanging out on Twitter and on indie hackers and talking. They're in an echo chamber in a bubble. And so that's why they write off their skills. But if they go out in the real world, just go to church or a friend's house and talk to people about this, they don't know anything about this. They don't even know what an AI agent is. Like, I go hunting every November with some buddies from my MBA program and they're all about my age. They're not really on Twitter, but they're working professionals, white collar guys making good money. And I talk, I'll talk to them about crypto or AI or whatever. And they're like, it's way over their head. Like you'd think they're 86 years old, living in, you know, Topeka, Kansas. No offense to anyone in Topeka, Kansas, but they just don't know. Twitter is a bubble. Like tech is a bubble.

I think people underestimate how early it is. It is just so, so, so early. We're 0.1% of the way.

There is N8N as valuable today as it was a couple months ago before this update.

I'll leave it to the watchers to decide. I think if you are a really technical person, like an engineer basically is who N8NK does to. And I'm not throwing shade at them, this is where they have their roots and I think this is awesome. It's like they're like an open source product and all of that stuff. But I think if you are either like not an engineer, like you're not that technical, or you are, and we have a lot of people in our customer base who are like, I can use other more technical products, but I just don't wanna. I just wanna solve a problem, you know, I think you are in need for a less technical, easier to use solution and that's what we focus on.

Yeah. I'm not about to say that N8N is gonna be hurting or going out of business like they're on a tidal wave. Right? AI is a tidal wave and there's a lot of room for everyone. But I look at it as like zapier and make. Right. Zapier is user friendly. Make is not. Make is cheap, Zapier is expensive. It's more for technical people. And zapier is not. I look at it like that. I know a guy that his whole, like his whole business is using N8N and like hundreds of customized complex N8N templates to help people start AI automation agencies. And, you know, so he'll train someone to go into a plumbing company and use N8N templates to, you know, be like a virtual receptionist. Virtual. I just don't see, and maybe I don't understand N8N well enough. I don't see why N8N would be necessary for something like that when you can just, you could just prompt it in Lindy.

Yeah, yeah, yeah. And not only that, you know, not only can you just prompt it with Lindy, like once you've prompted it, the flow editor that we show you is still 100x easier to use than N8N. If you want to set up an AI agent with N8N, you have to also set it up with a memory. You have to tell it what day it is. And the memory options that you are given include Cute, Rent, Vector Store, which again speaks to how much of a technical solution that is. Whereas with Lindy, it's literally just you click the AI agent button and you enter your prompt and that's it.

If I were to help a plumbing business or a small business, you know, a hundred thousand to ten million in revenue, automate some of their processes, right? Clocking in, clucking out software stuff, HR stuff, email, answering the phones. Is there anything that you can think of that N8N like you have to use something like N8N for that. Lindy 3.0 would not work for.

No, we've never run into a case where we were like, ah, you can't do this. You should probably use another solution. Look, at the end of the day, this is like a nerdy term that engineers use. And I said that as an engineer myself. All of these platforms are Turing complete. And Turing completeness is basically this concept of like, it's a universal computer. It's a universal platform that can do anything that any other computer and Turing complete universal platform can do. All of these platforms are Turing complete, including us, including Anita, and including everything else. No, you can definitely do the same things. I think it's just way easier to do it with Lindy.

Okay, forget the plumbing example. Is there anything that Lindy 3.0 can't do that you really need N8N for like anything complex, something that's just too technical for Lindy.

I can't think of a single thing. No, again, it's really more about. Let me think. They have certain features like mcp, right. That we don't support yet. And you would need to hack your way around. You can set up MCPS in Lindi, but it's not as friendly. And so it's not like you can't do it. It's like there are certain things that very technical users want that are easier to do in other solutions and explain.

To the listeners as simply as you can what an MCP is.

An MCP is the equivalent of a USB port for AI agents. It's just a universal connector. It's a protocol to connect AI agents with tools. And your tools being like your Gmail.

Not an API, but for AI, sure. And it allows you to have like an LLM on Like locally, right, on your own device or server?

I wouldn't say that. Like, it's USB port. Like, yes, if you want, you can use it for your own LLM. Like, but like, you can also have your own LLM on your own server and not have mcp.

That's a game changer. Because the people that I've spoken to that want to start an AI automation business, they're getting tripped up by NN because they're not technical. I'm not technical. That's not a drag on them, right? They're just not technical. And so it's like, how can I teach a plumber how to do this when I can't do this? But I feel like this removes that friction.

No. 100%. Yeah. And I think that again, that's the whole thesis behind the company, right? It's like AI agents are awesome and you shouldn't have to be an engineer to leverage them. Business users should be able to build AI agents because they have the business knowledge and that's where the value lies, right? And so if you have that layer of indirection between the guy who has the business knowledge on the one hand and the guy who has the technical knowledge on the other hand, now the iteration loop is so much longer because every time you want to make a change, you got to go through the engineer and it's going to take a day to answer to your email and it's going to take a day to make the change, and three days later you're going to have your change. It's not exactly what you wanted. So if you want to make five changes, it's going to take you a month. Whereas if you, as the person who has the business context, can make these changes yourself just by prompting, then you can have that iteration loop again. You can have your solution in that afternoon.

Oh, man, I'm just thinking of websites, you know, like, how long were websites in, like the old age, like in the 90s and 2000s, where you had to code, you had to know HTML to build a website and then, you know, tools like Squarespace and WIX came out and it made it easier, but there's still a lot of people making a lot of money building, you know, websites for others on Squarespace, Wix, WordPress, Card, whatever. And then now we're to this iteration of websites today where you just prompt it, you just prompt it in a Vibe coding tool or in Lindy and you get a website. I feel like we're in like that second stage of AI agents today that websites were in, like 10 years ago.

That's the perfect analogy. Like, imagine if, like every time, and a lot of people are still stuck in this world today, like, every time you want to make any change to your website, you've got to go through an engineer versus using Squarespace, Framer, Webflow, whatever. Or you could just go and change the thing yourself. So that's exactly the same idea. Yeah.

Well, is there anything about Lindy or anything cool that you're seeing that I should have asked you but didn't?

Let's talk about computer use real quick, because we talked about the agent builder, which I do think is like, the most exciting update. I am also really passionate about computer use because again. Oh, yeah, until computer use, you were limited by the number of integrations that you supported. And that's why Anthropic came up with mcp, because they're like, we want to make it easier for people to build these integrations and plug them into AI agents. But at the end of the day, you're still going to need an engineer at HubSpot or Salesforce or whatever to sit down and build an MCP instead. With computer use, you can have your AI agent use a computer and do anything that you can do on a computer, because a lot of solutions, they don't want you to use them through an API. They are never going to build an integration, let alone an MCP integration, because they don't want agents to use them. So I'll give you this example of I have a lot of spammers on Twitter in my mentions. It's really annoying. So I've set up this agent with just a prompt and I was like, hey, I want you every day to look at my mentions on Twitter and block the spammers. You can see every day it wakes up, it starts its computer, and then you can see on the right, this is its computer. This is what it sees and this is what it does. So you can see here, it's like it's found the spammer. I don't know why these people always tell me. It's like, this gentleman is awesome. Fire emoji. So annoying. And so she just goes and blocks, blocks, blocks, blocks. It's awesome.

Wow.

And again, it was just a prompt. And like, Twitter is not going to give you. Actually, famously, when Elon bought Twitter, like, they shut down the expensive. Yeah, yeah. So we don't need an API anymore. You know, we'll just clicking around and same on LinkedIn. I think I have another example here of, you know, these, like, comment magnets on LinkedIn that are really effective, by the way, as a lead generation mechanism. So, for example, five days ago we posted this tweet, this comment on this post on LinkedIn. And it's like, I built this AI agent. This is what it does. It's awesome. Comment agent. And I'll DM you the details. Check this out. 3,300 comments, responding agents. So when we had that, we're like, oh, this is a pain. We need to answer to like 3,300 people on LinkedIn. And like, you know, because every time we need, like, agent, hi just sent you a DM agent, you know, and in some cases we can send DMs, and in some cases people have their DMs locked down on Twitter, on LinkedIn. And so what we've done is that we've built this AI agent here. And what it did is like, okay, I found someone who commented. So step one, I'm going to open the profile and I'm going to try to message the person. So you see, like, it tries to click message. And then LinkedIn tells you, Ah, sorry. Like, this is actually sort of messed up because if you lock down your messages on LinkedIn, LinkedIn still less people message you, but only if they pay. You know, it's kind of crazy. And so you can see like here Lindy is like, ah, snap, I can't message them. So I'm going to go back to the comment and I'm going to reply in the comment. And you can see here, she's going in the comment and she's sending this message here, this comment that's like, hi, we'd love to send you the template. Please DM me. And this is just running on autopilot. And this is like, look, this is like 3300 comments. So even if it only takes one minute to do per comments, which I think it takes more than that, that is 55 hours right here saved, just not having to go and respond to those manually.

Flo, I'm paying like $350 a month for ManyChat, which just does this. It's $350 a month. And I use it on Instagram where it's like, if you want the business plan, comment plan, it's a lot of money. I didn't realize I could replace that with this. Like, I'm just now learning this. And then like, there's a lot of things that I want to be doing, but I need to hire real people for it. For instance, I get thousands of comments per day on Instagram and Facebook. And many of them are, are critical in bad faith, right? I look at like, negative comments as like, good faith. Like, hey, is that even realistic to make this much money? That's a good faith comment or bad faith? Like, you're an idiot. I hate you. You need to die. Right? And like, I just, I don't have the bandwidth to hire a virtual assistant to go in there and block them, but I'm not doing it. So they're just living there and it just bugs me. You know, I could totally do this. I could say, hey, like, discern the sentiment analysis from this comment. If it's a bad faith negative comment, delete the comment and block the person. And then like over a month, they'll just be so much more curated my comments.

It's funny, the example agent I just showed you of the Lindy that blocks Spammles and my mentions, after I recorded this demo I just showed, I actually updated it to do exactly what you just said, which is because, same as you, I get these people that are just like, rude. It's just like gratuitously rude. And so I updated the prompt to be like all the block mentions containing very rude or insulting content, especially when devoid of substance. And so, yeah, I don't. I no longer see these rude messages. It's like it saved my mental health. Honestly, even a little bit of followers on Twitter, like, people don't understand. Like, your experience changes. It just sucks. Yeah, you get these roles in your mentions. I no longer have that anymore.

Oh, that's so good. And like, Even tools like Manychat that, that have APIs to do stuff like this, like, there are automations that can respond to my DMS or that can respond to comments, Facebook, meta, Instagram, like, it will, it will block me. Like, it can harm my account because it thinks there's automated behavior or whatever. Like, I'm not asking you to like absolve me of any liability, but I feel like it doing it in an actual window makes it more like I have an actual employee doing it themselves and I'm safer from stuff like that.

That's exactly right. 100%. It's funny, actually, there is an episode of south park like that where it's like Cartman gets a lot of hate messages on Twitter and he asks but to like, like, filter them for me and but gets like brain damage looking at the head.

Yeah, that's funny.

Yeah.

Oh, man. I mean, even like people in my DMs, like, they'll send me three paragraphs, like, Dude, I started this business because of you. Like, I thank you so much. Like, you. And I don't see them. Sometimes I'll go in and look and I'm like, dude, this is amazing. But I don't have anyone checking. Like, I could reach out to them and be like, dude, thank you so much. It is AI, but it's better than not getting a response.

Okay, let me. Let me show you. I have another agent that does exactly that, by the way. This is going to look staged. Like, this is. We're literally just reefing for, like, the listeners here. But like, this. Literally, I have this agent here that's like my. Our ex. New follower.

I'm, like, softballing you all these questions. Oh, I think my viewers, they're used to this.

Those are pretty universal problems that businesses run into. Right? And so you can see every time we get a new follower, we have this Lindy, wake up. And she looks at the person who just started to follow us and she sends a message. So right here appreciates the follow. What repetitive stuff are you dealing with that's eating up your time? And we're actually working on enriching this agent right now. So, like, instead of sending always the same message, it can actually research you online first and be like, wow, I just saw your business did X, Y and Z. This is so cool. Hey, have you thought about using agents for this? And this can be tailored, man.

Yeah, well, dude, I think what I want everyone listening or watching this to know is these problems that I'm describing, you know, getting all these comments and DMs, it's probably not yours, but this is a problem that you can solve for other people that are publishing content. Right? Like, you can offer this as a service for people that don't want to read all these toxic comments or people that want to kindly respond to people in their DMs, maybe with a sales pitch, maybe with a thank you. This made my day. Like, this is a service you could offer for a monthly fee, just using Lindy on the back end.

That's exactly right. Yep.

You could use Lindy to find the customers, and then you could use Lindy to fulfill the customers.

That's exactly right. Yeah. And we have a lot of agencies in our customers that are, like, very successful at doing that.

Oh, man. This is. Oh. I don't usually get a headache within 39 minutes of talking, but today might be a record. And that's good. Headaches are good.

It's a brave new world. People don't understand what's happening? The capabilities that we're talking about today are literally 30 days old. So this is what I mean when I say it's so very early. I think it's funny because when I started this company, not that long ago, it was like two and a half years ago, I said, oh, I think we're not very far from having companies that are fully autonomously managed by AI agents. And people were looking at me like I just smoked crack. And no, it's happening. Like you're starting to see it.

Well, just think of like cold email automation agencies or cold email agencies. That's a thing. It's been a thing for years, right? I think that is basically to the point today where if I had a cold email agency, like my customers are companies that want to send cold emails, I could use cold emails to find my customers. I could use Lindy to do that. And then once I find them, I could use Lindy to craft, scrape, verify, send, reply to, reply to again, reply to again, close stripe API, collect payment. You could already do this today.

Yep, today. Like, this is not speculation. This is not a drill. This is happening 100%. What I really want someone to do is, yeah, I'm looking forward to see the first fully autonomous AI organizations set up by one guy. To your point, you could imagine the recruiting example. It's like, hey, these are three candidates. This is a docusign. You don't pay me anything unless I actually get you a hire through the door. By the way, an interesting thing is some recruiting companies work like that. It's on a contingency basis. You only pay us 20% of the first year salary, which is a lot of money. It's $30,000.

That's a lot of money.

But then there is fraud. It's like customers hire someone and then they will tell you and whatnot. So here what you can do. You can have another AI agent that checks every month the LinkedIn profiles of the candidates that you introduced to customers to see if they joined the customer. And if so, you hit up the customer and you're like, hey, where's my money?

Oh, so good.

You can set up the whole thing like that, end to end, like, hey, here's the docusign. And then once you send it, once the person signs it, like, now you have another agent that goes and starts finding candidates, reaching out to them, booking interviews with the customer, or sending them to customer to confirm first. And you can have the entire flow completely end to end, managed by AI agents.

I'm just picturing a company, getting an email, said, hey, remember when I introduced you to John Smith for that technical role? What happened to him? Oh, yeah, John, he passed. He took another job.

Really?

Because I see on LinkedIn he's joined your company. Invoice attached. Please pay within seven days. Invoice paid.

Yeah, yeah, yeah. I love that. That's exactly the idea. If not sent to Collections, and maybe Collections is another AI agent, you know, that's like, I'm sure it will be. This is a lawyer AI agent, you know, that's like, hey, you're in violation of.

Yeah, yeah. It's interesting because we're so early, but just like there are businesses that don't have websites today. We're going to be 10 years from now and there's still going to be people doing it the old way. Right. So that's not a reason for someone to wait, but it just takes a long time for people to adapt to these new technologies. So the sooner you act, the more you'll be ahead of your peers.

Yeah. I think at some point there is a tipping point where it's like, at first you get an advantage if you adopt these technologies first, and then at some point you are at a disadvantage if you haven't adopted these technologies today. People who aren't using computers and they're still using fax machines or whatever, I don't think there's a whole lot of them. If you're in that boat, you're no longer at an advantage if you use computers, you're just behind and you're at a disadvantage if you don't use them.

Yeah, I'm picturing like a spectrum. It's like, first phase is like advantage, first movers. Right. The biggest middle phase is table stakes. Right. You don't have a website, dude, that's table stakes. Yeah, like, that's just like. That's like you have to have a website. And then after that it's. You don't have a website. How are you still in business?

Yeah, no, exactly.

How do people know how to contact you?

Yeah, we're still in the phase of, like, advantage because, like I just said, like, this technology is like 30 days old. Like, AI agents started to work basically 12 months ago. Really? So it's still very early, which is why we are seeing the companies and the individuals and the agencies who are adopting these technologies first. Just completely blow it out of the water.

Oh, Flo, this is amazing. Lindy. AI Lindy.

Yep. Yeah, we're active on Twitter, we're active on LinkedIn. If you follow us on Twitter like an agent will be in touch with you.

Yeah, he's honest, folks. He's honest. Okay, if you like this episode with the Lindy founder flow, we spoke back on June 16, 2025, so you can scroll back and listen to and or watch that one as well. And as always, I would really appreciate it if you shared with a friend and used my Lindy affiliate link in the show notes. Thanks a lot. We'll see you next time on the Kerner Office.