Subscribe
Share
Share
Embed
Welcome to our podcast, where we dive into everything Go High Levelβfrom mastering the basics to tackling the most complex tasks. I use GHL daily in my business and rely on Google NotebookLM to stay ahead of the curve, keeping up with all the latest GHL features, tools, and innovations. This podcast is powered by AI, fueled by the research and insights I personally curate to bring you the most valuable and up-to-date content.
Copy this link for a free trial of Go High Level - https://www.gohighlevel.com/highlevel-bootcamp?fp_ref=amplifi-technologies12
Imagine landing this massive, like $5,000 a month client, right? You build out their entire lead routing automation, the ads go live, and then on day one, a single misplaced comma in some custom JavaScript, just completely breaks the data flow. Oh, that is the worst. Just brutal. Right. I mean, hundreds of high intent leads just get dumped into a digital void, and you are left scrambling to figure out why the system crashed. It's is uh, well, it's the ultimate digital marketing agency nightmare. So, welcome to the deep dive. We are thrilled to have you with us today. Absolutely. And we've got a great one for you today. We really do. And before we unpack how a new update is going to completely bulletproof your workflows against that exact uh, nightmare scenario, we actually have a massive perk waiting just for you. Right now, you can get a free 30-day Go High Level trial. Which is huge, by the way. Yeah, it's double the standard trial length. And the link to claim it is sitting right there in the show notes below. Definitely go click that and take advantage of it. It's honestly the perfect time to leverage a trial like that, especially considering the sheer technical power that has just been handed to agency owners. Exactly. So today's mission is all about pulling back the curtain on this massive update inside Go High Level. We've been, you know, analyzing the High Level support portal, their feature tutorials, the release notes, just to unpack this brand new feature called Code with AI. Yeah, and it's currently in beta for Go High Level workflows. Right, beta. And this update is designed specifically for you, the digital marketing agency owner, to build these really smart, highly complex automations without needing to know how to write a single line of code. Which is wild. I mean, we are looking at a fundamental shift in how automation logic is constructed here. Historically, custom logic has been the ultimate bottleneck for marketing agencies trying to scale up. Oh, totally. Like, if a client needed something just slightly outside the native capabilities of the visual workflow builder, the agency basically hit a brick wall. Okay, let's unpack this, because to really appreciate the solution here, we kind of have to look at the mechanics of the previous workflow behavior, like, the dark ages. Yeah, the dark ages of coding. Right. Before this beta roll out, if you wanted a highly specific, customized action, say, I don't know, calculating a tiered commission split dynamically inside a workflow. You had to write raw JavaScript from scratch. Right, within the custom code action. And writing JavaScript from scratch isn't just like a minor hurdle for a marketing team. It requires a totally different operational mindset. You're stepping out of visual logic and into syntax. Exactly. You had to worry about declaring variables correctly, managing data types, ensuring your code could handle unexpected null values without crashing the entire workflow node. It's a lot. And the sources point out that writing the core logic was really only half the battle. The real nightmare was the manual property mapping. Oh, yeah. The mapping was painful. Like, if a user filled out a highly specific form field in step one of your workflow, you couldn't just drag and drop that field into your JavaScript in step four. No, you had to manually write the exact syntax to pull that specific data point from the system's memory into your custom code. Right. Which means if you messed up one little thing, The whole thing breaks. That mapping process is where the majority of errors occurred, honestly. You were essentially blind typing the internal data keys. Wow. Yeah, if you missed a single bracket or, you know, capitalized a letter that should've been lowercase in the variable name, the connection severed. The workflow would fire, the custom code would run, but it would be operating on empty data. Which meant rigorous, tedious, manual testing. You'd have to push dummy leads through the workflow, check the execution logs, find the line of code that failed, rewrite it, and run another dummy lead. It was a massive time sink. Just hours gone. But according to the support portal, that old way of doing things is being replaced by AI-assisted code generation. Now, you literally just describe the desired functionality in plain English, and the AI generates the JavaScript snippet. It completely abstracts the syntax layer away from the user, which is amazing. It's like, uh, building a car engine from scratch every time you want to go to the grocery store versus just sitting in the back seat and telling a chauffeur where you want to go. That's a perfect analogy. Instead of playing this high stakes game of telephone where you hope a freelance developer understands your marketing strategy, this is like having a universal translator built directly into the platform. Right. You speak marketing strategy. For example, you type, calculate a 15% discount if the lead is a returning customer. And the AI instantly translates that strategic content into flawless JavaScript. Okay, I want to get into the actual interface here, because the chauffeur analogy, the universal translator, it all sounds great, but how does an agency owner actually interact with this inside the platform without, you know, crashing the car? Well, the user interface integration is designed to be totally frictionless. Based on the documentation, you just navigate to your workflow and add a custom code action, exactly like you would've done previously. Okay, simple enough. Right, but instead of being dropped into a blank code editor where you have to stare at a blinking cursor, you're presented with a new button labeled, Code with AI. Or in certain areas, it says Build with AI. So clicking that just opens up the assistant. Exactly. It opens a prompt window. And this is where you input your plain English description of the task. Once you clearly define what the code needs to do, you hit generate, and the AI engine processes your request to craft the necessary JavaScript snippet right there in the window. But wait, let's talk about the mechanism behind the automated mapping. Because you mentioned earlier how painful it was to manually write the syntax to pull input properties from previous steps. Oh, yeah, the automate property integration. Yeah. How is the AI actually pulling that off? Like, how does it know what data I'm talking about when I just use plain English? So, the AI isn't just generating code in a vacuum, right? It actually has contextual awareness of your specific workflow's environment. When you ask it to operate on a piece of data, the AI scans the JSON payloads, you know, the structured data tree from the preceding steps in your workflow. Oh, wow. So it reads what happened before. Exactly. It understands the available variables, automatically bridges the gap between the visual interface and the raw code, and it writes the proper mapping syntax into the JavaScript itself. Okay, I have to throw a massive red flag up right here. Go for it. The FAQs state that absolutely no JavaScript knowledge is required to use this tool. But if I'm an agency owner with zero coding experience, aren't we just handing a loaded gun to marketers who don't understand the logic they're executing? That's a very fair point. Like, if the AI hallucinates a command that sends an API request into an infinite loop, the marketer won't know how to stop it, let alone read the code to know it's just total gibberish. Yeah, that is the most critical question to ask when you're lowering technical barriers like this. But the protection mechanism lies in how the feature is actually deployed. Okay, how so? First, the generated code runs within a sandbox. That's a restricted environment within the custom code node, meaning it doesn't have systemic access to override core platform functions. Okay, that's good to know. Right. And second, the platform employs a very intentional one-click implementation process. The AI doesn't just auto-deploy its code. It presents the snippet to you first. But if I can't read the snippet, presenting it to me doesn't really help, does it? Well, it helps because of the iterative refinement process. The AI output actually includes comments explaining what the code is doing step-by-step in plain English. Oh, I see. Yeah, so if the explanation doesn't match your intent, or if the code fails the build-in test run, you don't have to try to manually edit the JavaScript. You just hit the regenerate button, or even more effectively, you refine your prompt. So prompt engineering kind of becomes the new coding. Exactly that. The documentation specifically highlights that the more precise your prompt, the better the AI generated code. Makes sense. They use a very revealing example regarding date formats. If you just type, fix the date, the AI has to guess what format the date is currently in and what format you actually want it changed to. And it might guess wrong. Exactly. But if you type, I have an API that returns a date in M-M-D-D-Y-Y-Y-Y format, convert it to Y-Y-Y-Y-M-M-D-D. Well, now the AI has the exact parsing parameters it needs to write definitive JavaScript. That makes total sense. Yeah. You're shifting your skill set from knowing how to write syntax to knowing how to clearly articulate business logic. Which is what agency owners are already good at. Right. And when we look at how that skill shift translates to the bottom line, the sources give us a literal playbook. There are five actionable use cases detailed in the documentation. What's fascinating here is how these technical features translate into immediate ROI. Exactly. Let's group these together because this is where the theoretical capabilities turn into actual practical takeaways. Let's start with what I call taming the chaos, which is data formatting and string manipulation. Oh, yeah. Unstructured data is essentially the dark matter of digital marketing. It really is. It's everywhere. It takes up massive amounts of space, but it's entirely unusable until you process it. Think about how leads come in from a dozen different sources. You have phone numbers with dashes, some with parentheses, some with country codes. And some that are just a messy string of 10 digits. Exactly. And if that data isn't standardized, your automated text campaigns fail. The system simply won't send an SMS if the phone number field isn't formatted properly. So, instead of spending hours manually cleaning CSV files, you just use Code with AI to generate a standardization script. You tell the assistant to parse all incoming phone numbers and reformat them into your agency's preferred standard before the data ever even hits your CRM. That is huge. And then string manipulation takes that a step further. We've all seen those automated notification emails from third-party lead providers where the actual lead information is just buried inside this messy unstructured paragraph of text. Yes, those are the worst. You can't natively map a paragraph into a first name or email field in your database. Right. And historically, extracting that data required writing regular expressions, or regex, which, let's be honest, regex is notoriously difficult even for season developers. Oh, Regex looks like a string of random punctuation marks. It is incredibly easy to break. But now you just provide the AI with the plain English instruction. You say, scan this block of text, find the sequence that matches an email address format, pull it out and save it as a distinct variable. It acts as a translator, turning chaotic paragraph payloads into neatly structured CRM data. Which naturally leads to the next phase. Once your data is perfectly clean and structured, you usually want to send it somewhere or use it to fetch more information. Which brings us to API integration. Escaping the walled garden. Yes. We are talking about generating HTTP request code to seamlessly fetch data from external services that, you know, maybe don't have a native Go High Level integration. Because native integrations are fantastic, but they will always be limited by business partnerships and development cycles. Like, if your client is a specialized contractor, they might use a highly obscure, industry-specific database for their materials pricing. There definitely won't be a native app for that. Exactly. But with this feature, you just instruct the workflow to build the HTTP request. You can say, take this zip code from our clean data, ping this specific external weather API, and check if it has hailed in that zip code in the last 24 hours. Wait, really? That specific? Um, yeah. And if it has, you automatically trigger a roofer's outreach campaign to that exact neighborhood. That is mind blowing. You're basically building custom micro apps that run seamlessly inside your workflows. And the AI handles the complex part of APIs, which is uh, properly formatting the headers, securely passing the authentication tokens, and handling the JSON response that comes back. Right. So you've pulled this highly specific external data into the workflow. What do you do with it next? Well, that's where the final two use cases come in, mathematical calculations and conditional logic, and they combine to create this hyper personalized routing. We all know how to build nested if/else branches based on simple tags. Like, if the lead has the VIP tag, send them down path A, if not, path B. Right. Pretty standard stuff. But standard visual logic breaks down when the routing depends on dynamic, real-time variables. The sources talk about computing interest rates or dynamic discount percentages. Let's look at a real scenario. Say you have a workflow for a financial services client, right? Okay, yeah. A lead comes in, your new AI-built API integration fetches their real-time credit score from an external service, and now you need to calculate a custom interest rate offer based on a complex formula that factors in their score, their stated income, and the current base rate. A standard visual workflow simply cannot run that kind of algebraic formula on the fly. No way. But the Code with AI feature can generate the script to take those three variables, run the calculation in milliseconds, and output a highly personalized interest rate. And then, the conditional logic use case takes over. Instead of routing based on a static tag, you tell the AI to route the lead based on the result of that complex math. Like, if the calculated interest rate is below 5% and the external API confirms they are a homeowner, route this lead directly to the senior broker's calendar. It allows you to program highly customized, multi-variable decision-making paths. You are no longer constrained by the limits of a visual builder. You're really only constrained by your ability to articulate the strategy. Here's where it gets really interesting though. It's like having an invisible, microscopic data scientist living inside your agency's workflows, instantly adapting to whatever task you throw at it. That's a great way to put it. But let me pump the brakes for a second. We've painted a picture of absolute workflow Nirvana here. But we need a reality check before people jump in. This raises an important question for sure. High Level's own documentation is very clear about the guard rails. Before anyone listening goes out and attempts to, you know, rebuild their entire agency infrastructure this afternoon, we have to talk about the limitations. Don't go firing your developers just yet. Right. The first and most obvious parameter is location. This capability does not currently exist everywhere in the platform. Oh. It is specifically housed within the custom code action inside workflows right now. You cannot summon this AI assistant in the email builder or the funnel builder to write front-end code. Good to know. And the second major caveat is that the feature is explicitly tagged as beta. Which is basically software terminology for, "This is powerful, but it's actively learning." The underlying models are still being refined based on edge cases and user feedback. And because it's in beta, the documentation strongly emphasizes the human verification element. Right. The FAQ section asks point blank, "Will the AI generated code work perfectly every time?" And their answer is a very transparent no. Yeah, they're honest about it. While the AI is engineered to generate functional scripts, especially for complex operations like parsing API responses, it's highly recommended to manually review and thoroughly test the output before deploying it in a live, client-facing workflow. You absolutely must utilize the testing features within the workflow builder. Run test contacts through the node, check the execution logs, and ensure the output matches your expected business logic. The platform grants you full ability to manually edit and modify the generated code, meaning you maintain total oversight. If we connect this to the bigger picture, the human is still the architect. You have to know the structural integrity of the campaign you're building. How the data needs to flow from room to room and what the ultimate conversion goal is. Exactly. The AI is simply the builder, pouring the concrete of the JavaScript based strictly on your blueprints. If your blueprint is flawed, the building will still collapse. That is a perfect distinction. Strategy cannot be automated. Only execution can. So, what does this all mean? If we pull all of these pieces together from escaping manual property mapping to generating complex string manipulation and dynamic math, the core value proposition here is that Go High Level has fundamentally democratized the execution of complex custom code. They really have. They've removed the steep learning curve of JavaScript syntax, allowing agency owners to build infinitely smarter, more capable automations with significantly less friction. Zero manual coding needed. It completely changes the primary question agency owners ask themselves. You no longer have to ask, do we have the technical resources to build this? You only have to ask, what is the most strategic thing we should build next? Right. And actually, looking at the trajectory of this beta, I think there is a highly provocative final thought we need to consider. Oh, let's hear it. Well, we've established the mechanism here, right? An AI model can read the contextual data of your workflow, understand your plain English intent, and instantly write functional logic to bridge the gap. It is already deeply integrated into the architecture of the system. Right, we talked about how it knows the JSON payloads inside and out. Exactly. So, if the AI possesses that level of systemic awareness, how long until the workflows themselves transition from being reactive to proactive? Wait, what do you mean? Think about it. How long until the system is actively analyzing your agency's performance data in the background, noticing an inefficiency in your lead routing, and autonomously writing and suggesting a custom code optimization without you ever even typing a prompt? Oh, wow. Workflows that autonomously audit and rewrite their own logic based on real-time conversion data. That sounds like science fiction. But given the automated property mapping we just discussed, the foundational mechanism is literally already there. The jump from prompted generation to autonomous optimization is much smaller than the jump from manual coding to prompted generation. That is wild. That gives you plenty to think about as you look at your own agency's automation architecture. And before you go, I want to speak directly to you, the listener, one last time. Do not forget the massive puck we discussed at the top of the deep dive. The link for your free 30-day Go High Level trial is sitting right there in the show notes. It's honestly the ideal sandbox. There's no better way to understand the power of AI-assisted coding than by testing those exact five use cases we covered today. Absolutely. Go click that link right now, take full advantage of the double-length 30-day trial, jump into a test workflow and try building a custom API integration or a dynamic math equation using just plain English. Step into the role of the architect, let the AI act as your builder, and see exactly what you can create. Thank you so much for joining us, and we'll see you on the next deep dive.