Transcription
Hey, I'm JB and on today's video, we are going to give you a presentation that I am calling advanced AI for beginners. I was invited to speak to a small group of entrepreneurs about AI. And my goal was to open their minds about what's now possible. And on today's video, I hope to do the same for you.
So jumping right into this advanced AI for beginners is not necessarily for beginners but it's for maybe chat GBT users or probably better to be power chat GBT users the ones who think that they're automating their business by utilizing chat GBT. Now my goal is to take you way above that where you can truly automate tasks within your business and just open your mind to what's possible. How can you replace a bunch of different tedious systems or repetitive systems that happen in your business?
So, first I want to talk about automation in the old way and automation with AI. Automation in the old way was if this then that. And I think the most simplistic automation that people are probably really really familiar with is Gmail automations. So in Gmail you say if this email comes from this person then label it as this but it's super logical there's no decision making there.
Now what AI does is it says if this happens so if new email is received then analyze that email then make a decision and then actually do whatever you have now decided to do. So in this scenario, instead of having a filter for every single email that comes through your inbox, now you can look at it and say, if any email comes in, then read it and determine out of a predetermined set of labels that I already have set up inside of Gmail, which one it would be most relevant to. So hopefully you can organize your entire inbox by doing something like that. And we're going to show you probably 5 to 10 different more examples on how we can do things the old way versus the new way.
This is just a quick visual that I probably should have scrolled to a little bit earlier, but take a quick look at this. New email received from Joe. Apply important label. New email received. Now, OpenAI reads the email and then not only provides a label, but can also have the ability to now draft a response that you're all programming in through Zapier or Make.com or NAD.
I will also take a quick minute to give Zapier a huge shout out. Zapier has been something that has already opened our eyes and ears to automation now, but that's always been an extremely logical way. So we personally as a marketing agency have automated so many things in our business just through Zapier. But then what happens is a human has to really be involved here. And so now with the integration of AI nodes within that process, you're really able to allow it to make a lot more decisions.
Put on your thinking cap here and get into this AI automation brain. Think about how any system within your business can be automated without human intervention. And think about all of the different tasks in your business that are repetitive that happen over and over and over and over again. Even if it does provide creativity or decision-m that's okay. If it's something that you do a lot, let's try to think about that right now. And that way you can, you know, kind of reference that list. If you want to pause this, take a minute to write down maybe five different systems in your businesses or tasks in your business that can be automated systems.
I'm going to talk about human in the loop versus replacing human decisions. You may, especially if you've already gone down this AI automation rabbit hole, you may have heard the term human in the loop. And that is going to be really, really important. But we don't want to use that as our crutch. We want to try to replace this human in this scenario. And when I say that, it's not to fire people. It's to allow humans to just be more productive and actually have oversight over these AI systems.
A really good example of where human in the loop is really important is if you know your AI agent is doing something on a browser and then it needs to log into your Gmail and that browser is currently logged out and it doesn't have your password then it's going to require you to put in your password. Or another good example is I have AI many times draft responses to emails. Well, this is still me and my reputation, my identity, and and there's limits to how much I want AI to just replace who I am as a person. So, I want it to draft the response, but then I want it to let me know that it drafted a response. So, that way I can go approve it or edit it or do something along those lines.
Now, where I think many times people say human in the loop is they automate certain things and then have human decisionmaking that happens like read this document. Oh, I'm going to create this document and then have it proofread that document and do this this and that and then then send it back into the AI. Instead, I would encourage you to replace that proofreading. So have it create the document, then have a different agent that's separate from the one who created the document as if a proofreader and have the types of things that you would look for in that document. So is it too botlike? Do you wanted to make it more human? Do you need to fact check it on a few things? So every time it states a statistic, have it go look for that statistic online and try to provide a source to it. And even though it's all AI, in some cases need to think about AI as being individual people with individual goals. And there's ways inside of these different automation platforms that you can create different agents that don't know what the last one did. All they have is what's in front of them right now, their goal and their prompt, and then they're going to review it in a different way. So you could ask one AI to create a logo, but then you can have another AI receive that logo, not know who that other guy is, make critique to the logo, and then send it back and forth in that way until the reviewer is is happy.
So now we're going to go through a series of different things that hopefully inspire you to think about old automation versus new automation. We try to make these as relevant to as many people as possible. There's many common issues that every single business and business owner has. We tried to create examples that we feel like would resonate with as many people as possible.
One is lead followup. Every company hopefully has leads and they need to follow up with it. The old way is if someone fills out a form, send them a templated email to everyone. So the moment you fill out a form, you say, "Thank you for responding. We'll get back to you shortly." The new way is if someone fills out a form, AI reads their message and sends them a personalized message response based on their specific inquiry, tone, and urgency. So, it may get the same exact message across. But if somebody spent a lot of time putting in a good description, it could not only just respond, but it could actually crawl that person's domain, which is typically their business website, and then determine what that business does, determine who their target audience is, what services they provide, and then it knows what you're trying to sell is essentially. And then maybe it kind of amps them up, gets them excited about it. Hey, I know that you offer consulting services to CPAs. Our AI will help you do that so much quicker because of this, this, and this. Whatever that scenario is, try to think about how you can improve that experience.
Next one is phone support. And this is actually kind of a funny example that we have with one of our own clients right now is we have a voice AI that is capable of handling sales, service, and parts within a car dealership. But this client only wants us to be handling sales cuz he has an automated service scheduler. So he says, "Press service or press one for service, go this way. Press two for sales, go this way." So that's a little bit of an old mindset. The new mindset is let the AI just answer the phone call. Let's get rid of the IVR. Have it say, "Hey, this is Summer at ABC car dealership. How may I help you today?" And then if it says service and that client still wants to send it to the service scheduler, then let's go ahead and do a live transfer into service. Now, if it goes into sales, then boom, we got this. Our goal is to schedule an appointment or be as helpful as possible and answer as many questions as possible. So, think about how it can replace not only humans, but just old technology like IVRs.
Invoice categorization. So, I will say we haven't actually built this because luckily invoice categorization or what I'm about to mention is built into a lot of the account payable software that we have right now, but it easily could be something that you build as well with a PDF reader, an AI powered PDF reader. So, the old way is if an invoice comes from QuickBooks, mark it as accounting. The new way is AI reads every incoming invoice, extracts the vendor amount and category, and logs it appropriately in your finance tracker, even if it's a PDF from a random vendor. So, it's able to scrape things. Scraping is not AI. And I think that this is becoming a common misconception because scraping is becoming more common because it empowers AI to do better because the more information AI has. But I hear so many people say, "Oh, I built an AI powered web scraper." And it's like, no, web scraping has just always been here. But where it becomes AI is when you scrape a PDF or a website and then you make an analysis on it. So, it could scrape everything off of a PDF, but then AI really has to determine, oh, this segment of information, this segment of text is the invoice number. This over here is the client's or the company's logo. And does it need image recognition to pull the name out of that and determine who the true vendor is? And you know, these are different line items cuz every single invoice is going to be designed a little bit differently.
Next one, cold outreach. Before we used to upload a lead list and then send the same cold email to a thousand contacts. The new way is AI research is each lead's website or LinkedIn profile generates a fully personalized cold email that speaks to their business model and auto sends via email or LinkedIn.
Okay, so example here. So cold email, we are probably all familiar with mail merges. We merge in somebody's first name, put it in the subject line and then the subject is John is this you? question mark and then that creates curiosity and and that's still a very valuable strategy. And then under that then in the email body it might mention that company's name or whatever that may be. We don't really need to do all that anymore. I mean we're still going to customize things, but now we can actually download a list of 10,000 records, scrape the URL, put that into a database of a summary of that person's website. You can download a list of contacts, have it scrape the website, then have it determine who is that website's target audience or who is that company's target audience and what do they provide. Then write a cold email like I mentioned earlier in this video that says, "Hey, we see that you do this or we've been really impressed." Always add compliments inside these cold emails. Hey, we are really, really impressed about how you provide websites to dentist offices. What we want to do is make your websites better by improving the process with this AI system that we're trying to sell. So now what I would encourage you to do is think about that cold email as a mail merge. So now you go to your traditional email sending platforms like Mailchimp or whatever you're using your CRM. Now upload your data and then within that data you actually have your cold email lines in there. So you don't even have to write your cold email inside of Mailchimp anymore and put different mail merges in there. You just have first name is the merge, intro line, and sales line and structure your email in a templateed way where all the data already includes all the different cold emails.
Another way that this is extremely impactful is I know for us we used to download certain lists like you know especially if we're targeting car dealerships for example we used to download Acura then send out an Acura email and then Honda and send out a Honda email. Well, now we can just upload every single brand and then have it talk to that brand however we want to talk about it. So, we might talk to a Mitsubishi dealership different than we would talk to a Mercedes-Benz dealership. So, it can actually look at that brand, then write the cold email based on the personality of that and then say we have done this with this many of those brands. You can even that's still kind of niche in car dealerships, but let's just say it's B2B companies and you have CFOs and CPAs and consultants and you know marketing agencies and all this other stuff in here. You can sell your service and ask the AI how it would sell its service to that particular company literally just within Google Sheets or Google AI studio or you know depending on how complex you want to get. Reason why I recommend other ones is cuz Google Sheets has sort of limits. I think it can only load 200 about 200 AI formulas at a time. And to be honest, that's probably okay cuz it's pretty quick. And then depending on your Google Workspace account, I think it limits you to like a,000 or 2,000 per day. So then you literally have to come back another day or have other users populate with their own credits. So it can get a little timeconsuming. It's not totally easy to do in bulk.
All right. Missed calls or demos. So old way if someone misses a scheduled call send sorry we missed your call in an email. The new way is the AI just calls the lead back reschedules the meeting based on your availability and sends a calendar invite all with you without lifting a finger. So the moment that you miss a call which really you shouldn't be missing calls cuz AI should be picking up your calls. So either whether it's AI picking up your call right away or AI picking up right before your voicemail hits there should not be voicemails anymore. AI can completely replace voicemails, have a much better conversation, and still take a message.
Okay, so how do we do this? If you've watched our videos, you already know how. If you are new to this channel, then we are going to show you the three different operating systems that can power all of these different types of automations. One is Zapier AI, second one is make.com, and the last one is N8N. So, we've created a comparison chart to show you which one might be best for you. Let's just summarize this really quick and then we'll go line by line. But Zapier is the easiest, but also most limited. Make.com is still easy, drag and drop, very visually appealing, a lot less limiting if you're okay with a slightly higher level of complexity. And then naden is kind of complex. So, you either probably should be a developer or be able to hire developers and to work inside of NADEN. I will say don't let NAD fool you. It says it's a no code platform and hypothetically maybe you could make do without code, but you're going to want to know how to code if you're inside of NANE or else things get really confusing and daunting really really quickly. And then also the power of NAD is being able to customize it with your own code. So if you don't want any sort of code, go to make.com or Zapier.
So ease of use, we've talked about this a little bit. Zapier is simple, drag and drop. Make.com is really visual but slightly more complex. And then naden has a steeper learning curve in terms of customizations. Zapier is really really limited. But I will also say just jumping down to integrations, they have a lot more out ofthe-box integrations. So when I say limited, they're not crazy crazy limited. Um, but they can only really do certain functions that they have in there. And then make.com supports a lot more complex logic. So I'd almost say like start in Zapier and then you'll see where you outgrow certain things and then go into something like make.com and then if you just know, hey, I'm building a business and I have money to invest or time to invest in this or I'm a developer and I know I can commit my time to this, then it's fully programmable. You will not be limited by NAM. We have done the craziest stuff already and every single idea that I throw at our team, they just do it. It's almost like just traditional coding in some ways. I know a lot of traditional coders are moving to naden cuz it's easier to code inside of a platform like this that provides a little bit more structure.
Scalability. Zapier is really great for simple workflows. Zapier even doesn't trigger right away. Sometimes they trigger like every couple minutes or every 5 minutes. Make.com is better for medium complexity and naden is ideal for complex and scalable automation. So it is a lot quicker. You know, like I said, I would build a business on top of naden. I might be hesitant to build a full piece of software on top of Zapier or Make.com.
Developer friendliness, we've already really spoken about that, but Zapier does have a little bit of scripting. Um, not too much though. Most of it's just plug-and-play things. make.com leverages web hooks and API integrations. So you can actually integrate to more things than just the thousand apps that they have in here via web hooks mostly. And then NAD is built for developers. Integrations out of the box for Zapier is massive. That's what they've always been known for. Make.com and NAD just expect that you may have to build custom integrations. Being an API developers is really critical for NAD. Uh, web hooks is something that somebody that's pretty simplistic can actually learn. So, if you want to build custom integrations inside of make.com, just look into web hooks. See if that's too complex for you or not. The last one's easy. Best for a non-technical users for Zapier, mid-level users for make.com and power users and developers for NAN.
Now, we're going to talk about the AI automation landscape. And to be honest, I'm not going to spend a lot of time on this, but I have a concept that this industry map that we created helps you visualize a little bit better. So, this is by no means a comprehensive list of AI tools. But what I wanted to say with this is many people, especially business owners, say, "We want to have AI automation inside of our business." So, then they buy one tool or they buy seven tools, then they feel defeated because they're like, "Oh, that's not really automating anything in my business." They might have a video creation tool or a content writing tool or all these things that may be making them more efficient, but they're not truly building automated systems. They're just giving their team more tools that do things faster.
Now, where true AI automation, in my opinion, uh when it comes to building automated systems comes into play is taking multiple ones of these and piecing them all together. So, let's just say you're using make.com and then you want to say anytime we receive a sales lead. We want this LLM to read that email. Then we want it to determine what its response should be. Then we want to send that content over to Hey Genen and we want Hey GenE to to select an avatar and actually respond with our copy. Then we want to connect Gmail and embed that video inside of Gmail and send every single new sales lead 100% custom video avatar. Now, even that's pretty simplistic. You'll continue to think of other things along the ways of how now as a stakeholder I receive SMS alerts and all these different things that can happen, but really just wrap your head around the idea of finding all the different tools and then figure out how to connect them all. And it's not about figuring out how to. The how is NANE, make and Zapier. And there's even other ones out there, but those are the three primary ones. Try to figure out how you can connect all the dots and what's most valuable for you.
So, now that we've said all this, what ideas do you have? Please comment your ideas down below. And not just to provide another comment, but I will most likely actually just create content on it. So, if you want us to build it, then feel free to put your ideas down below. We'll map it out. Whether we're just mapping out and showing you how to build it or maybe we feel that it's valuable enough to just build just to show all the viewers on this channel. I want to do that. I want to provide that level of content that's triggered by you all. So please, please, please comment with your ideas down below.
Subscribe to my YouTube channel right now if you like this type of content and you want to build and promote and sell AI systems. Maybe you just want to build things to make your own business more efficient. Then this is the channel where you will learn a lot of that. And I try to keep things as simplistic as possible so you can feel inspired and not intimidated. So I hope to see you in the next video.