Automate Everything: Demystifying AI Agents with Jacob Bank

 
 
 
 
 
 
 
 
 

SHOWNOTES

This episode is a must-listen if you're a solopreneur or small business owner who's curious about AI but doesn’t speak “tech.”

I sat down with Jacob Bank, founder and CEO of Relay.app—a no-code platform that lets you build your own AI agents (even if you’ve never written a line of code). Jacob shares exactly how he uses 40+ AI agents to run marketing, support, sales, finance, and more... all without a team.

This isn’t theoretical. Jacob is doing this right now—running a full company solo, with the help of AI. 

What You’ll Learn in This Episode:

  • What an AI agent actually is (and why it's not just a chatbot)

  • The simple 5-step workflow Jacob uses for repurposing content on autopilot

  • The one chart that changed everything (hint: it's got 40 AI agents on it)

  • How to think in flowcharts instead of tech terms

  • Why AI is still underhyped—yes, underhyped

  • The critical skill that every solopreneur should start building now

This is one of those episodes you’ll return to whenever you’re ready to take your business automation to the next level—it’s that good.

RESOURCES:

[FREE LIVE TRAINING] Build Your First AI Agent with Relay.app

 
 

When I first discovered this tool last week, I immediately reached out the founder, Jacob Bank, to invite him on to my podcast and to invite him to deliver a live training for my subscribers (you!). He said yes to both! Jacob is going to walk us through how to build your very first AI agent from scratch. You’ll learn how to streamline your workflow, save hours every week, and get started with automation — no tech overwhelm, no coding required.

What You’ll Learn:

  • What an AI agent is and why it matters

  • Common patterns and use cases

  • How I'm personally using AI agents in real-world scenarios

  • A live walkthrough to build your first agent from scratch

Check out Relay.app here,

 
 

TRANSCRIPT

[00:00:01.740] – Gary Vee Hey, guys. It’s Gary Vaynerchuk. You’re listening to the Front Row Entrepreneur podcast with our girl, Jen. [00:00:10.020] – Jen Lehner Hey, everybody. Our guest today is the founder and CEO of Relay.app, a platform that makes it easy for everybody to create their own AI agents. Before that, he was the product lead of Gmail—like that Gmail—the founder and CEO of Timeful, and an AI researcher at Stanford. In other words, a real smarty. Welcome to the show, Jacob Bank. [00:00:38.900] – Jacob Bank Thank you so much, Jen. This is going to be fun.

Read more...

[00:00:41.060] – Jen Lehner I am so excited. Here’s how you got here. I was messing around with Relay.app, and honestly, I can’t remember how it crossed my desk, but it did. For my membership, I have an AI club where I help solopreneurs and small businesses leverage AI in their businesses. I screen all the stuff, pull out the best stuff, and then bring it to them. I’m looking at, I don’t know, many, many dozens of tools every week and experimenting with them. Then came your tool, and I was completely blown away. We’re going to talk about all the use cases and all of that. But I think the standout reason that got me so excited is that my audience—and I think most people—are not AI experts. They’re not techies. What you’ve done is create this very, very helpful tool that’s going to allow us to get so much more work done, and it’s plug-and-play, and it’s honestly dummy proof. It’s like, “Click here, click there, and away you go.” Then after I started playing with Relay, I went over to your LinkedIn, and I saw this post, and it was amazing. The post, I looked at it today, has 20,905 comments. [00:02:11.880] – Jen Lehner What? You don’t have a marketing team. Is that right? [00:02:15.740] – Jacob Bank I am the marketing team—me and my agents. Yeah, it’s so funny. Let me tell you a little bit of the backstory of how we got there. My background is in product and technology. I was an AI researcher, I was an engineer, then I was a product manager. I’ve always been a product builder. But for one reason or another, the way the Relay App Company has developed is we have a very strong team of engineers, designers, and a product manager. Those are eight people. They’re building the product, and they don’t need me to help build the product. What my job is at our company is to do every non-product function. I am the marketing team, the support team, the customer success team, the sales team, the finance team, the HR team, the recruiting team, and the operations team. Those are my nine functional responsibilities, and that’s all me. We have no additional contractors, no interns. I do not have an executive assistant. I do not have a VA off somewhere. It’s just me. The only way that I’ve been able to do this is I use AI a lot to help me get work done. [00:03:20.440] – Jacob Bank This post that really resonated with people on LinkedIn—I created a little org chart. But instead of that org chart being people that I’ve hired on my marketing team, it was just a little AI agent that I have working with me on our marketing team. I think that really resonated with people for a couple of reasons. One, because it showed where the future is going in terms of how businesses are going to be built. Then two, it gave people a lot of really concrete ideas of things they might want to use for themselves. I did have a hesitation about doing that post. Let me tell you the hesitation I had about the post, which was my whole mission is to communicate AI in a way that is simple, approachable, practical. I really want to demystify it. There are many AI communicators that I think are making it much more complex than it needs to be for a typical person. I’m sure you’ve seen those 200-node diagrams of N8N workflows, or you’ve seen those really complicated presentations about LLMs and vector databases and RAGs and knowledge bases and memory and JSON objects and all these technical terms. [00:04:29.600] – Jacob Bank I don’t think you need to know any of that to get value from AI right now. The one thing that gave me pause is that by putting this org chart of 40 agents up, it could come across as a little bit complex, but I wasn’t trying to show complexity there. I was trying to show opportunity and how cool this can be once you get good at it. But if you look at any individual one of those 40 agents, they are extremely simple. Let me give you one extremely simple example. I have an AI agent that helps me whenever I create a video: it automatically drafts a LinkedIn post for that YouTube video because it’s bread-and-butter content repurposing. If you’re putting all this effort into making something on YouTube, you might as well get value out of that content on LinkedIn, on Twitter, to your email newsletter, to your blog, etc. It’s a lot of work to do that content repurposing manually. Here’s all this AI agent does. When a new video is posted on our YouTube channel, it wakes up, it goes and gets the transcript of that video, it runs it through an AI prompt much like you would in ChatGPT that says, “Hey, based on this transcript, please write a LinkedIn post promoting this video. By the way, here are three other LinkedIn posts I’ve written before for similar purposes. Use those as examples, go fetch the thumbnail image, and then post it.” This is not complex. There’s no magic. If you can write down that five-step flowchart, you can build this, and it gets the post about 96% right every time. I usually have to tweak, and this is another thing that’s really important—I have a human in the loop for pretty much all of these workflows. Anytime content is created on my behalf, anytime an email is being sent to a customer, I’m always reviewing, editing, tweaking it much like I would if I had a junior content person working on our team. But it’s a very simple workflow and it saves me a lot of time. The most important thing I want to communicate is—and there’s two things I want to communicate. Number one: AI is underhyped, not overhyped. Underhyped. I don’t think we have all fully digested how much this is going to change the working world. I’m here in this weird little San Francisco AI bubble where people are debating the subtleties of different models. [00:06:36.700] – Jacob Bank But in the world as a whole, it’s way underhyped. We don’t realize how good it is yet. Second: it’s ready right now for everyone. You do not have to be a coder. You do not need to be a technical expert. If you can create a flowchart—on paper—if you can write, “First this happens, then this happens, then if it’s a client like this, do that; then if it’s a client like this, do that; then do this.” I think if you can create a flowchart, you can create an AI agent for yourself. [00:07:04.500] – Jen Lehner I love that so much. You’re speaking my language. I’ve been preaching the same thing, and that is why your post, I think, resonated so well. I’m going to share this, of course, you guys. It will be in the show notes, or even better, you can go to Jacob’s LinkedIn profile and connect with him over there. It’s Jacob Bank, B-A-N-K. He’s making bank right now with this new thing, I’m sure. I would hope so. But anyway, it’s all these very simple workflows that all of us can use because we all have to do marketing. I’m sure there are many more agents that you could get more specific with for any specific industry. But it’s just mouthwatering when you look at it. [00:07:48.500] – Jacob Bank Another reason I’m so excited about small businesses and solopreneurs in particular is I have never met a small business owner who said, “I’ve got so much extra time to do all the things I need to do.” It doesn’t come up. Small business owners are typically serving multiple functions within the business. They have a lot of different things on their plate. Many of them are distracted from the main reason they got into the business in the first place. And small businesses are not going to be able to go and hire a 10-person marketing team. That’s not even on the realistic menu of possibilities. And so the reason I’m so excited about this is for every artisanal jam maker who just wants to make artisanal jam and not worry about managing invoices and order forms and promotional posters and social media shorts, they don’t have to do any of that stuff anymore. They can have AI do it for them. [00:08:44.380] – Jen Lehner It’s so exciting. All right, well, we’ve used the term “agent” many times already in our conversation. Can you define for us what an agent is? [00:08:54.440] – Jacob Bank Yes. I first want to give you the technical definition. Then I’m going to tell you to ignore the technical definition in favor of a more practical definition. All right. The technical definition is: an autonomous software system that uses a large language model to achieve a goal by recursively calling a set of tools. Let me break that technical definition down. All it means is when you have a task that you need to accomplish, there’s some flowchart of the steps that go into that task. Let’s say your task is to get five new leads for your business. Your flowchart goes: first, look in the Yellow Pages to find the 100 fast-casual restaurants in our ZIP code; then find the email addresses of the managers of those; then send them a personalized email; then look at the replies you got; then fill in five. There’s some flowchart of a task. The technical definition of an AI agent is that you do not tell it the flowchart. You give it a goal—“Find me five leads”—and a set of tools—you can look up information on the web, you can make cold calls, you can update spreadsheets—and then the AI agent is responsible for turning that goal and that set of tools into a successful outcome. [00:10:11.500] – Jacob Bank That is the technical definition. I think the practical definition is going to veer further and further away from that technical definition. Here’s the practical definition: an AI agent does work on your behalf. This is different from a chatbot. You’ve probably used ChatGPT or Claude—one of the chatbot tools. A chatbot is an interactive experience, just like you’re text-messaging another person. You ask it questions, you have it do research for you, you have it do some tasks for you, but it’s always in the context of this interactive conversation. That’s a chatbot. Another way that we’re all interacting with AI is a copilot. Many people listening have probably used Grammarly before or a tool like Grammarly that whenever you’re writing an email or a document, it injects itself in and says, “Oh, you spelled this wrong,” or, “Oh, you use passive voice here,” or, “Oh, you used bad grammar.” That’s a copilot, because as you’re doing some other task, the copilot is helping you out. A chatbot—you talk to; a copilot—it helps you out. An agent, much like a human employee, does work on your behalf behind the scenes without you needing to directly interact with it. [00:11:20.660] – Jacob Bank You don’t need to directly chat with it. You don’t need to directly interact with it as a copilot. I think that’s the practical definition of an AI agent that people need to know. Whenever you think, “Oh, this AI is waking up on its own, it’s doing some work for me on its own, it might check in with me to get some guidance, but fundamentally, it’s autonomously completing the task,” that’s what an AI agent is. The reason AI agents are so exciting: they’re exciting for two reasons. One is, sure, it’s nice to have work done for us while we’re sleeping. It’s very, very powerful to have something with that level of capability and autonomy that it can really take work off of our plates rather than just be an in-the-moment copilot. The second thing that’s so exciting about AI agents is we already have a mental model for AI agents. I’m sure every single solopreneur or small business owner listening has the list of five tasks they would delegate to an intern or assistant or a contractor if they could. We all have that list where it’s like, “Oh, man, I spend a lot of time replying to comments on LinkedIn,” or, “Oof, I spend a lot of time looking through Facebook to find leads,” or, “Oof, I spend a lot of time parsing out information from our invoices and putting them into the spreadsheet.” [00:12:33.000] – Jacob Bank We already have a mental model that’s like, “Oh, yeah, I could hire a person to take that task off my plate and do it for me.” But we haven’t had the ability to do so either because we haven’t had the resources or it’s too hard to find the people or it’s too hard to train them. Now, those same exact tasks that you have had a person do, you can create your own AI agent to do them. [00:12:51.180] – Jen Lehner Okay, amazing. One thing that just popped into my head that I don’t want to forget to ask you that I am very curious about: at the speed at which all these platforms—ChatGPT, Claude, Genspark, all of them—are adding new features and new capabilities, is that good for Relay or bad? Let’s say ChatGPT Tasks, for example. [00:13:16.960] – Jacob Bank I’d say it’s more good than bad for us in the sense that things that are currently impossible in our product become possible as the underlying model providers get better. It is not totally obvious how the landscape is going to evolve. Is OpenAI going to decide to build an agent-building platform like Relay? Maybe. Even that, I would argue, is good for us. [00:13:40.870] – Jen Lehner Yeah. [00:13:41.520] – Jacob Bank Yeah. Just in general, in the startup ecosystem, starting startups don’t die because a bigger competitor decides to play in the same space. I’ve been in this market for 15 years. I’ve literally never seen that happen. Startups die because the founders run out of energy or run out of ideas, or they ultimately can’t build something that users like. If OpenAI announced tomorrow that they were building a product in our space, that would be the best news ever because it would bring so much visibility to our category in our space that, yeah, a bunch of people are going to use the OpenAI version. But the OpenAI version is also not going to work for people for various reasons. It might not have the integrations they need. It might not be easy enough to use. It might be missing specific capabilities. It might be too expensive. They’re going to come looking for an alternative and they might find us. Yes, AI models getting better is pretty much always a good thing for us. There may be opportunities where we have to simplify our product or change the way our product works because of AI improving in capabilities. [00:14:42.170] – Jacob Bank But no, this is just purely good news that these models are getting better. I’ll give you an example. I was helping someone out with a use case a few months ago where they got a PDF of a calendar from one of their suppliers, and they couldn’t convince their supplier to give it to them in an importable format. They just got this PDF. What they wanted to do was build a simple workflow that takes in that PDF, uses AI to automatically figure out the start time, the end time, and the title of each event, and then write that to a separate Google or Outlook calendar. None of the models six months ago were good enough to do that. It sounds simple. It’s the thing a human intern could easily do, but none of the models were good enough. Then I tried it again last week with Gemini 2.5 Pro, and it worked. It totally worked perfectly. I was like, “Wow, that is a thing that was impossible, that required a human.” That is not a fun job for a human—to look through a PDF of 50 events and manually transcribe them. I want humans to be doing creative, emotionally resonant work, work that builds on their relationships and skills. [00:15:48.460] – Jacob Bank I do not want people to have to be copying 50 events off of a PDF into a digital calendar. Let’s let AI do that for us. [00:15:55.680] – Jen Lehner Okay. And let’s talk about how you were walking the talk, okay? Because I just want my listeners to know what I observed with my own eyes, which is why I had to have you on the show, because—and you guys will see this when you go and Google him and you go to Relay.app and you go to the YouTube—because all the things are going to link you to the YouTube trainings that are exceptional. I’m on your email list, so I’m getting emails of all these workshops that you’re doing, and the workshops that you’re doing are real, true workshops. You’ll sign up for one of these free workshops, and then Jacob will say, “Okay, click here. All right, everybody, have you done that? Okay, great. Now, for those of you who haven’t connected your Gmail yet, I’m going to wait for you to go ahead and do that.” Then he gives you a few seconds. You connect your Gmail, and then he’s going to walk you through a really cool workflow that 95% of people are going to love. That has something to do with LinkedIn and email and whatever else. And so what I’m saying is that here you are, you’re in charge of this whole amazing company. [00:16:56.220] – Jen Lehner You are the only one doing all the marketing, all of this stuff. So when you said you don’t want to look at 50 PDFs and extract information because you want to be doing this stuff that matters, that’s exactly what you’re doing. So you are living the example. [00:17:15.210] – Jacob Bank Yeah. So let me… Because if you look at the amount of content I produce, it’s three to four YouTube videos per week. It’s one to two emails to our newsletter. It’s four or five LinkedIn posts. It’s four or five—I probably do 100 direct customer support emails. I probably meet with 20 customers per week. That level of output. And it’s not because I’m superhuman or I’m working 100 hours a week. It’s like I have a really good system and really good tooling around it. Let me tell you how these workshops actually arise. What happens is I’ll get an email from a customer that’s like, “Oh, I’ve read it. It turns out to be a really interesting place for me to find leads for my business. It turns out a lot of people in the B2B marketing subreddit are perfect leads for me.” Or pick your favorite subreddit—Houston Real Estate or whatever. A lot of people in this subreddit are really good leads for me. How would I build a workflow that can automatically monitor that subreddit and tell me if there’s a conversation that’s relevant to my business? Let’s say I’m a real estate agent or a mortgage broker or whatever. [00:18:22.900] – Jacob Bank What I’ll do is I’ll reply to that email and say, “Oh, yeah, you could do something like this, like this, like this.” I’ll make a quick template. I’ll put the template in our gallery. I’ll go to my calendar and say, “Yeah, this would be a good live session. This would interest a bunch of people.” I’ll make a 30-minute event on our Relay.app events calendar that says, “Build with me: Reddit monitoring agent,” and then my system will take over. We’ll automatically get an email to our newsletter promoting that session. We’ll automatically get a LinkedIn post from my account promoting that session. We’ll automatically get a slide deck created with collateral for that session. We’ll automatically get a shot of the template I just created for the user for that session. And so all I need to do is understand what the customer is trying to do, what they’re struggling with. I need to show up for 30 minutes to teach people how to build it, and the rest just happens. And now—well, again, there’s a human in the loop. I’m always checking the email to make sure it’s good. I’m checking the LinkedIn post to make sure it’s good. [00:19:20.520] – Jacob Bank But as you know from creating content, so much of the effort is just the activation energy of getting that first draft. And once you have a first draft that’s decent, it takes five minutes to get it over the line. But when you have that need to write a LinkedIn post hanging over your head and you don’t have any ideas and you don’t have a draft, it’s excruciating. And so, yeah, there’s no magic to it. I listen to a customer, I hear what they’re struggling with. I think it would be useful to other people. I create a 30-minute calendar event. That’s all there is to it. And all I have to do is show up and build the thing. [00:19:54.050] – Jen Lehner Yeah. And then, and meanwhile, you’ve got all these templates in Relay that we can just press the button and then all we have to do is just color by number. [00:20:05.580] – Jacob Bank I will give a caveat on the templates, though. Templates, I think, are a double-edged sword. Templates are great for providing inspiration of what’s possible. They’re great at giving you ideas of what might be valuable for you. My templates—I don’t think they’re quite paint by numbers because the subreddits I want to follow will be different than the subreddits you want to follow. The exact content I’m looking for to know if it’s a lead for my business will be different than whether it’s a lead for your business. I might want the summary sent to me over Slack. You might want it in a Google Doc or sent to you as an email. There is enough customization that you do actually need to build up some skills here. This is really, aside from Relay.app, this is my mission. I think that the ability to build your own AI workflows and agents is going to be the most important professional skill in the next decade for basically everyone. Certainly every knowledge worker sitting in front of a computer, certainly every solopreneur, but even also every real-world small business owner. We’ve got cupcake bakers, we’ve got contractors, we’ve got chemical manufacturers. [00:21:19.480] – Jacob Bank Everyone has work that they need to be using this for. And yes, you could pay $20,000 for an off-the-shelf solution that will partially solve your problem. But really, you need to build the skills to be able to create your own. And so that’s all I’m trying to do. I’m trying to deliver. And the really cool thing is now I don’t have to do any marketing for Relay. All of our marketing is just I create educational materials, and some subset of the people who find those educational materials useful will sign up for Relay.app. Some subset of those people will pay for a subscription, and then the flywheel returns. I’m trying to produce content in every format that makes sense. I have my 30-minute or one-hour YouTube build-alongs. I’ve got my five quicker YouTube videos. I’ve got blog posts. I’ve got templates. We’re just about to kick off a five-day email course on AI Agent 101 basics for small business owners. It’s funny. That’s all I’m trying to do. I’m just trying to create as much valuable education as possible. [00:22:14.420] – Jen Lehner It’s so smart because I never quite understood why there’d be great products out there, then content creators will create a library of content of their own about said product. I mean, Mailchimp, ConvertKit, whatever it is—totally. They’ll have some lame and dated obligatory training materials on their website. But I just think it’s such a missed opportunity. I’m sure there are plenty of creators out there doing their thing. [00:22:46.900] – Jacob Bank The creator ecosystem is going to be important to us because I can create content about things I know about. I can create general content on, “Here’s how to make a trigger. Here’s how to make an action. Here’s how to run a prompt.” I can create good general content. I can create good content about the use cases that I use in our business—for marketing, for support, for lead generation, etc. But I don’t know what a commercial real estate person does all day. [00:23:45.290] – Jen Lehner Wow. Do elaborate. [00:23:47.220] – Jacob Bank Here’s an example. I run webinars once a week, sometimes twice a week, sometimes three times a week, just depending on when I get inspired. But at least once a week, I typically get between 50 and 400 people showing up live, and then another 1,000 to 2,000 that watched the recording afterwards. I already told you my process. My process is I come up with an idea, I make an event on the calendar, I show up, and I build the thing live. I don’t do any prep because part of the fun of attending one of my live sessions is I make mistakes all the time. We all make mistakes all the time—like, “Oh, man, I can figure that filter wrong,” or, “Oh, shoot, I used the wrong step type there,” or, “Oh, I didn’t write that perfectly in my prompt.” And this has double benefit. One is it saves me a lot on prep time. Number two, it’s way more useful to the audience because no one benefits from seeing a perfect demo where everything goes perfectly, because that never happens in real life. That’s my webinar process. I put in probably five minutes of prep. I show up, I do five minutes of follow-up, and it serves a couple thousand people. [00:24:52.840] – Jacob Bank I’ve participated in a few webinar series from larger companies in this space. I’m not going to name any names; I don’t want to make anyone look bad. But I’m participating in larger webinars. The way a typical larger company with 500 or 1,000 or 2,000 employees works is they have multiple people in their marketing team. They have a newsletter person; they have a webinar content moderator; they have the product marketer who’s responsible for talking about the features. And so for this webinar I did, there were four pre-meetings. We had to record the whole thing. It took six weeks of prep to get it done. And guess how many people were at the webinar? [00:25:32.200] – Jen Lehner Twenty-nine. Fifty? [00:25:33.200] – Jacob Bank Wow. Fifty. And then another 500 watched later. So I’m not saying the webinar was a failure, but I’m saying all that work and all those hours and all that bureaucracy and negotiation—that was just unnecessary. It was just unnecessary. I would way rather take one super-motivated individual who’s empowered, who has all the contacts, give them some AI tools to help them, and they can outproduce a team of 10 easily. [00:26:00.460] – Jen Lehner I love that. Earlier, you were talking about the skills that we need… That your templates aren’t really exactly color-by-number. Although I will say that they’re color-by-number enough, because I think the hardest thing for many of us is to—and that’s why your, back to that org chart that you did, is so valuable is because it allows us to think of scenarios like, “Oh, I didn’t even think that that was possible.” [00:26:29.100] – Jacob Bank Exactly. There are three things that are hard about creating an AI agent. I don’t know which is the hardest, but they’re all hard. The first is: what problem do I have that is appropriate for an AI agent to solve? Where do I even start? That’s where I think the org chart is really valuable because it gives you 40 starting points. Number two: What is the flowchart of that task? How would I instruct a person on my team—a contractor or an intern—to do that task? For those of us who studied logic, studied computer science, creating flowcharts is very second nature because we’ve studied this for 20 years. But creating flowcharts is not actually an easy thing to do. It really takes a certain structured way of thinking. I’ll give you a funny story. I remember whenever you take an introductory computer science class—it’s teaching you a structured way of thinking—the teacher will always have an apple core in their hand and say, “Tell me how to throw away this apple.” Then the people in the class will say, “Oh, just walk over to the trash can and throw away the apple.” And he just won’t move. Then they’ll say, “Turn left 90 degrees.” He’ll turn left 90 degrees. “Oh, cool. Now lower your right arm 90 degrees.” He lowers it 90 degrees. “Now drop the apple in the trash can.” He drops it next to the trash can. And so what you learn is that when you’re working with a computer, you have to tell it exactly everything you want to do: move your right hand 90 degrees up, rotate your body 90 degrees to the left, take two paces forward, look down, bend at the hip, and look down. And similar, I was talking to a customer a few days ago, and they basically said, “I just want to get stuff from Monday.com into Google Calendar.” And I’m like, “Okay, that’s a good starting point. We know Monday.com is the beginning and Google Calendar is the end—but I need more information. What do you want to get at the end in Google Calendar?” “Oh, I want an event that corresponds to an installation of some blinds.” “Great. And what happens in Monday.com that makes you realize you need that event?” “Oh, when I add a new item to the Monday.com board, I want a new event in Google Calendar.” “Okay, great. Do all Monday.com items from this board always get a new calendar event?” “No, only the ones that are moving to the stage ‘Installation Ready.’” “Okay, great. Do they all go to the same Google calendar?” “No, of course not. There are eight different Google calendars for the eight different installers.” “Great. Where does the start date come from? Does it come from this field in Monday.com?” “No, not that field—that’s the due date field. It comes from this other field.” You eventually get into this flowchart that says: When a new item is added to this particular board in Monday.com and the field is set to “Installation Ready,” look up the installer column to select the right Google calendar to add it to, then add the event to the Google calendar using this title, this start time, this end time. [00:29:31.300] – Jacob Bank Constructing that flowchart—it’s not a technical skill, but it’s very important. It’s a very important skill to build. Then the third challenge is, once you’ve got the workflow in your mind, actually translating it into Relay.app. That’s where we have a lot of work to do to make the product better, to make it easier for people to do that. But you do need some basic technical skills. You need to know what a trigger is—a trigger is what wakes up your AI agent and tells it to start doing work on your behalf. You need to know what an automated step is—an automated step is something that does an action, like sending an email or creating a calendar event. You need to know how to write some instructions for AI. I’m not going to use the word “prompt” because that’s a technical term—a prompt is just instructions that you give to AI, just like you give instructions to a person who’s coming to help you with something at your house. And so what I’ve been trying to do is create educational content at each of those levels. [00:30:24.720] – Jacob Bank At the top level: how do you think about what AI is good at and where it may fit in your business? At the second level: how do you learn how to break down a specific problem in your business into a flowchart? And then at the third level: how do you take that flowchart and actually build it? And most of my “build with me” sessions are on the third component. [00:30:42.720] – Jen Lehner Okay. So let’s say you build it. You’ve figured that out. You’re good to go. You’ve done your Monday— which, for those of you who don’t know, Monday.com is a project management platform like Trello or one of those. And then, okay, so you’ve got your stuff going from Monday over to Google Calendar. That’s what she wanted, right? Yeah. Okay, so all that’s good. Yay for us. But then, how do you make the agent predictable and reliable? [00:31:10.040] – Jacob Bank Yeah. So there’s two answers to this. The first answer is: when you have anything high-stakes, I recommend having a human in the loop. And that means you double-check. So, for example, in this Monday.com example, you might want to have a quick human-in-the-loop step that says, “Hey, we just saw a new job for Jen that needs to be scheduled at this address at this time. We’re going to put it on Jacob’s calendar because he’s the installer. Is that correct? Should I do that?” “Yes, go ahead and do that.” “Oh, no, shoot—I forgot to mark that Jacob is out of office that week, so switch it to Mike, the other installer.” One way to make it trustworthy is to have a human in the loop. The second way to make it trustworthy is you’ve got to test it—test it regularly. Because even if you get this thing right the first time, it’ll drift. I promise you, it’ll drift because one installer will leave your team, another installer will join your team, the calendar mapping will be a little bit off, or maybe you’ll change your installation convention or the way you write addresses. [00:32:20.080] – Jacob Bank While the most important point I want to communicate is that everyone can do this and that it’s crazy valuable and it’s super simple, I also want to communicate that these are not “set it and forget it.” This is another reason why I think skill building is so important, because I have so many customers that come to me asking, “Can I just pay someone to build this thing for me?” Yes, you can pay one of our partners to build it for you. Yes, you could pay one of our partners to maintain it every time it breaks. But just like learning how to work with a spreadsheet, Microsoft Excel was a valuable career investment for many of us—I think it’s worth it. I think it’s worth it. I really encourage everyone, no matter how senior you are, no matter how much you think you’re not a technology person, you can do it. You can learn it, and it’s going to be worth it. [00:33:08.440] – Jen Lehner Okay, well, I just found our sound bite for all of our promos right there. So thank you, Rick. Because yes, we’re just going to—I like that—and put a big exclamation point on it. I love that. [00:33:17.700] – Jacob Bank Getting back to my point about AI being underhyped, this is not optional. I’m telling you, I’m in it, I’m using it. This is not something you get to be skeptical about. This is not something you get to say, “AI is overhyped. It’s not going to matter for me or my business,” or “They don’t get me or they don’t get my business.” You just can’t operate that way. I’m telling you, this is not that technology trend. This is the technology trend that is going to change every single business. You can either be one of the people who’s using it to your advantage, or you cannot. [00:33:48.020] – Jen Lehner I love that. Well, on that note, looking ahead, what are you really super excited about in the next few months? [00:33:55.940] – Jacob Bank So all three of those problems I’m excited about solving. So number one, I’m excited about creating way more inspirational material of where AI can be useful in a wide variety of businesses. For example, I just recorded a video a couple of days ago on AI for residential real estate agents, which is not something I knew anything about—but I helped a couple of our customers out, and I came up with, “Oh, wow, this is a super-cool use case.” Every time a new listing comes in from the MLS, we can automatically look at our database of clients that are actively looking. If the new listing meets their criteria, send them a personalized email saying, “Ken, I have the perfect four-bedroom, two-bath for you in this neighborhood. It’s squarely in the middle of your price range. It meets all your criteria. I think you can get it.” I just love learning about new use cases and then turning those into templates, turning that into content, turning that into partnerships. Super excited about that. Then second, I really want us to continue to make it easier to get what’s in your brain into our tool, because I appreciate your kind words about how easy it is. [00:35:00.740] – Jacob Bank We’ve worked hard to make it easy. I do think we’re the easiest product in the market right now, but I want to be 10× easier. I think we can be 10× easier. I think we can get you to the point where all you have to say is, “I want to get some of the things from Monday.com into Google Calendar,” and then just with a natural text-based interaction, we’ll get you all the way to a working workflow. [00:35:19.860] – Jen Lehner I love that. I think you could do it. I do. [00:35:22.480] – Jacob Bank It’s hard, but I think we’re actually close. I think we’re close. I think six months from now, we will completely open up who can benefit from AI. It’s exciting. It’s exciting for me because I think I told you, I always had this personal dream that I was going to be a teacher one day, and that was always what my career was building toward. For a variety of reasons, it ended up not happening. I ended up being a grad student, then an entrepreneur, then a product leader. But now I get to be a teacher again. And so really, all I do all day is figure out things that I think would be valuable to lots of people, and then figure out how to communicate that in a YouTube video, in a blog post, in a LinkedIn post, and then see what feedback people have and go from there. So it’s really fun and rewarding for me. [00:36:07.260] – Jen Lehner That makes so much sense that you had that background because I think that’s what it is. And that’s what drew me to you: you really are such a good teacher. And of course, the product is so great. So, listeners, Jacob is offering something really super generous to just you guys, just the people listening: 500 extra AI credits. And I want to say your free plan is super generous by itself, because you can actually—unlike a lot of these platforms—they actually give you enough to work with to set something up and get a feel for it and see how it works before you have to upgrade. But on top of that, Jacob is offering us 500 extra AI credits. But what you have to do is you have to send an email to jacob@relay.app, and I’ll put all these in the show notes, and put in the subject line: “Front Row AI Credits,” because as we discussed, he’s a very busy guy. [00:37:08.940] – Jacob Bank And then I’ll figure out how to create a little AI agent that takes care of that for me. [00:37:13.410] – Jen Lehner Perfect. I love it. And those of you in the Front Row AI Club, you’re in for a treat because Jacob’s going to actually be coming and doing a training for us later on in July. So I’m super excited. And you have a lot of work you got to go do. So thank you so much for spending time with me today, Jacob, and I hope we can have you back sometime. [00:37:33.770] – Jacob Bank I’d love to. Thanks so much, Jen. This is super fun, and I love to do live trainings anytime for any reason. That’s my whole thing. And so if anyone else in the audience is interested in running a training for their team or their local community or their business, just send me an email, jacob@relay.app. [00:37:48.370] – Jen Lehner Oh, I love that. All right. Awesome. Thank you, Jacob. Thanks so much.

 
 
 
 
 

jen lehner

Cart (0)