How to actually make money with AI in 2026

2025 was the year that everybody talked about AI, but 2026 is going to be the year that businesses actually have to go about implementing it. And so what I'm going to do in this video is talk you through exactly what that implementation process is going to look like, especially for small to medium businesses, and then what needs to actually be done and how you can get involved in the trillion dollar transformation that's going to happen to the businesses that need to essentially do a lot of reorganization to adopt AI and be able to succeed in 2026 and beyond.

Now, before I jump into all of that, really quickly, my name's Adam. I've done over 60 projects on Upwork. I'm in the top 1% program and their expert vetted program. Uh so I specialize in freelancing. I have a freelancing core and I use those skills to go ahead and build a global low code development studio uh which builds internal systems for businesses similar to what I'll talk about in this video. As well as that, I also teach uh freelancing and AI skills in the AI builder hub and I obviously make a lot of YouTube videos on AI coding frameworks and small to medium business adoption of AI.

Now, with all that out the way, I'm just going to jump in here and really just talk you through exactly what 2026 is going to look like when we talk about small to medium businesses adopting AI and exactly how you can get involved in taking a slice of that pie for yourself if you can get your skills in order and figure out where you can slot into this puzzle to start making some money and start helping businesses generate real ROI with AI implementation.

Okay, so this is our diagram. And we're going to start here and talk about the AI coding money map for 2026. So what I'm going to do is talk about how small and medium businesses and pretty much even larger business but especially small and medium businesses are set up in 2025 and what's needs to happen for them to adopt AI because what we saw this year was in 2025 there was a lot of hype a lot of people talking about AI but actual implementation into businesses outside of a couple flashy LinkedIn posts was actually quite minimal. We had a lot of people talking about voice AI. We had a lot of people talking about, you know, all sorts of cool agents, but actual implementation really lags behind. And so there's a couple of reasons for that, uh, that I'm going to talk about in a little bit more detail in a moment.

Um, but when I talk about small to medium businesses, so small to medium businesses, I talk about typically businesses that have anywhere from 5 to 100 employees. Now, everyone kind of defines this differently. So we've got the startup stage, really early stage. I'm going to focus mainly on like small to medium enterprise or businesses. And typically these companies have two setups. So I'm going to put this here. And they have two options really. So they either have a number of existing tools that they use to run their business. So I'm going to I call these SAS tools. Uh and this this operates every single part of their operations. So they've either got SAS tools or they've got some sort of like manual process. So I'm going to drag these two and give you an example in a moment. But depending on the industry, it's a bit wonky.

So if we talk about SAS tools here, uh, let me just grab a circle and I'll just draw them in here. So typical SAS tools that these businesses are using is they'll have they'll have some sort of CRM. So we're going to talk about CRM. Now typical CRM like HubSpot, HubSpot, Salesforce, Pipe Drive. Now these tools are typically to store their clients. So like if we take a real estate company as an example, uh, they will store a lot of information about their their clients or even a gym, the serum typically stores all of your paying subscribers and your team members. So that's that's one tool that you would use to run your business. Um, because if you didn't do that, then like you know 20 years ago, you'd have to write all of these down on paper and then if someone quits, you have to have someone that's manually constantly updating these forms. If we have them all in a tool like that, we we can access that data online. um, we can send emails. So there's people we can we can kind of connect all these things. Um, then again you typically have like file storage uh or they call this like a CMS a content management system in here like notion koda google drive uh etc. All the Dropbox all those kind of tools u that's managing like your information uh documents proposals all that kind of stuff. Uh a lot of businesses will have some sort of like document signing tool for example uh for contracts employment contracts proposals that kind of stuff. So we're going to tool like Panda do Panda docuine etc. Um, then they also have project management tools. So like if you take a software development agency for example which is what we do you have a project come in a client come in through some sort of lead channel. Uh so that will typically go into your CRM. So let's say someone wants to build an internal system with me and they come through our landing page and through that process we then store them in in a CRM and we set up all these automations to automatically message them to try and get them to jump on a discovery call. Then they jump on a discovery call and uh from there you know we have to do a number of things. We have to qualify them. We have to build a proposal. We have to do some research and an AI audit to uh basically you know audit their company, build financial projections and ROI projections for them. There's all these things we have to do. We typically manage that on a project management tool. Uh so these typically look like Trello, Asana, Jira are just a couple to name a few. Jira. Um, then you've got your communications. So depending on what kind of industry you're in, a lot of professional services will use something like Slack, Microsoft Teams. U some some people use WhatsApp if it's if it's a little bit more casual. So you'll have those kind of applications. So you can see right there we've got that's five different tools in itself. And then every business is niche. So construction tools will have or construction platforms will have uh specific apps for quoting and takeoffs. Um, real estate companies will have specific platforms that they'll use to you know understand micro prices and do research. So this can go to like five or 10 different tools and there's so that that's that's one model of how businesses are operating today.

Uh, now option two is the more manual processes approach. Now some businesses are not this setup uh because in today's day and age like people think that this is organization having everything on the web in the cloud super put together. Now it can be if you have a if you have a a tech team for your business that's really on top of exactly how this works. It can plug together quite well but there's a lot of process automation involved to make this work. And I'm going to talk about exactly what that looks like with with AI in a moment. But option B is like manual processes. So these kind of businesses, they're using email a lot. Uh, so they're using typically they have fairly basic text stacks. They're using Google Drive, Google Sheets, a lot of CSV management, um, a lot of emails. So here there'll be a lot of emails, um, and like really simple tech stack, but a lot of manual overhead. They've got VAS that are managing things, uh, virtual assistants that are, you know, responding to a lot of emails. that they they've got all their kind of project management stuff set up in Google Google Sheets and Excel. It's personally not my favorite way of doing things. I think it's quite slow, quite janky, and takes a while to train other team members on that kind of information. Uh, but this is like the this typically more construction businesses, a little bit less technical. These are the two kinds of setups that we have in business today.

So, a lot of people when they try and grow, they get caught in a lot of the overheads that are with manual processes. They're hiring all these staff to do, you know, manually do payroll, for example. So, we haven't even there's also accounting tools in here. So, here you've got your accounting and you've also got payroll. Um, so here they're doing like payroll. They're doing it all manually. They've got all the different manual processes and this adds up to, you know, hiring 5, 10, 15 staff of 40 hours, 40 hours a week to do really manual, boring, repetitive tasks because a lot of business processes at the end of the day are actually extremely repetitive. And so what happens is companies do this for a while. They realize they've got all these overheads and they go, "How can we automate this? How can we make it cheaper? How can we make it easier for our team to stay on top of what's happening?" And so then they go down this approach. But the problem with this approach is you end up paying thousands of dollars a month for these tools. But even then, it's, you know, you you still got five to 10 different tools. Team members have to be onboarded and learn how to use every single one of them. You have to copy and paste all this data between each of these. And that in itself still has a cost. It can be it can be more uh economical than this approach, but this still isn't ideal.

And so the real opportunity in 2026 is from AI coding. And what's happening in 2026 is so this this is the 2025 scenario. So I'm just going to put this here. This is the 2025 and before is this. Let's make this a little bit bigger. Cool. So this is 2025 and before. Then we have the 2026 scenario. Now what's happened now is we have a suite of tools. So I'm just going to put this make this look like this. We have a suite of tools. Now these tools look like at an entry level. I'm going to put level one AI coding. Now I think this is these these tools are you can build complex applications in this and I'll explain I've explained in other videos why I think there's level one and level two but these kind of tools look like lovable bolt v 0. Now what's happened is in 2026 we've had AI coding come in which has significantly dropped the price and time it takes to build software. Now there's some pretty ridiculous statistics out there but you know n everyone everyone in the entire world is using technology for you know 25 to 50% of their day at a minimum and there's like something like.3% of the world is proficient in JavaScript and some of the basic web coding languages. Now, I can't remember the exact stats, but it's it's something ridiculous for the amount that everyone is is engaging with technology in in today's day and age. So, what these tools have done is they've seriously democratized and lowered the bar to entry for people to develop their own software. So that has it that this has serious ramifications longer term on the software industry in in general because the only reason now you can go check out my uh AI is killing SAS video if you want to understand a little bit more about why these these tools and this model of business is under serious attack in the next few years. But essentially it's it's to to do with the fact that AI is significantly reducing the cost and time it takes to develop software. So what's happening is we have these tools coming out that are essentially allowing people to b build their own internal business system. So instead of having to juggle 5 to 10 different tools. They can they can strip a good 80% of these tools out there. Uh and that that has a number of benefits for businesses. Um, but at the at the end of the day it's unifying all their data in one spot which is allowing them to adopt AI.

Now I'm going to talk through exactly what the opportunities are going to look like in a moment. Um, but I call level one AI coding bolt zero v 0. Um, and yeah, lovable bolt v 0. Level two AI coding. Uh, and this is for more complex applications. Looks like um some sort of AI coding framework. BMAD is the one that I', it's completely free and open source. It's the one that I've been using the most and it's personally the one that I think is the best at this current point in time.

Now, the what is the key difference between these two things? The key difference is documentation. Now, when you're building anything with AI and even when you're using like chat GPT or Claude and you're trying to generate proposals, documents, just ask it questions, you notice that you have to create separate chats. And what happens is the longer that you use a chat, the less accurate it gets because you've put all this extra information in there. And we call that like context. We call it context dilution. It's the same way that if I'm having a conversation with you about how to do something and then I pick up the phone, I ask what you want to eat for lunch today. I start asking all these random questions, it starts to dilute our conversation and so you start getting distracted by me talking about other things that are not related to the specific task at hand and it reduces the the clarity that you have on exactly what we're doing. And the same thing applies when we're using AI chats. So what happens is when you use these tools, they don't have a uh very clear open- source documentation process. So the the issue with that is that when you have a large complex application with all these different features, it starts to forget why it's built things in certain ways because it didn't document that process down. So it will build a certain app in a certain way. Um, an example is let's say you have like five different users in your app. So you have admins, you have um clients, you have users, whatever. You build it in a certain way. You prompt it. You tell it, hey, look, admins should only be able to see this admin portal. Clients should only be able to see the client portal and users should only be able to see the users portal. You you do that in a conversation. It builds all these things. But then a month later, you come back and you start making changes and it starts editing different portals because it's forgotten in the context because it's so diluted. It's forgotten why you segregated those pages in the first place. And so the difference between these two is that the documentation that's generated from this BMAD framework and these AI coding frameworks is set up in a way that it persists over time and you can access this information later down the line in different conversations. So it's essentially just lower context dilution. Now if you're building small projects that then you may not need this level of complexity. So this could really suit more basic style applications complex apps. I think this the level two is essential. If you want to learn more about that, you can dig dig a little bit deeper into the BMAD framework and how that works. Um, but that is the this is the general reason um and the major shakeup that's coming in 2026.

And so what's going to come out of this? Well, we have two we have two approaches for automating businesses in 2026. So we have actually I'm just going to put this here. So we have option one and option two. So option one text. So option one and I call this process automation. Process automation with existing stack. Now if you're someone that's in the software space a lot, you'll see tools like uh Salesforce, they're all coming out with custom agents. They're building AI agents that can do all sorts of cool things. And a lot of businesses, particularly enterprise, they won't, this is still too complex for certain enterprise clients that have all of these really, really strict security requirements and there's so much red tape around everything that building custom software just doesn't really make sense for them. And so what happens is they turn to process automation. So essentially what they do is they use on a smaller level um what happens is there's automation tools. Now, automation tools like make.com, make.com, Zapia, uh, and NAN, essentially in most cases, what these tools are doing is they're building workflows that connect these tools together and aggregate the data, uh, to then go ahead and adopt AI, build AI automation, stuff like that. Because what what happens and what I should actually describe before I do this, sorry, is what's going to happen in 2026 is the rise of what I call the context problem. What what is the context problem? The context problem is the fact that to build So let's take three let me quickly build three basic AI automations that are really really popular um to explain this. So cuz what what do we actually mean when we say AI adoption? Let's build three examples. So let's do one uh outbound. Let's outbound cold call uh AI voice. So this that's one that's one option for an implementation of AI where we can set up AI agents that have real human voices that can make outbound sales calls and cold calls to potential clients without needing a sales agent. So that's one example of AI implementation that could have serious ROI for a business. Now let's do number two. Let's say automated automated project scoping/p proposals. So option two is using AI instead of needing to use a expensive human who typically has a lot of domain experience. So if you want to code out a complex software application for example to be able to scope that out correctly to understand the tech stack to know exactly what milestones to put in there to understand how long certain things will take to require someone with a lot of domain expertise uh and they are typically expensive because it takes a lot of experience to you have to have done it many many times to be able to digest a project and build a build a detailed scope of proposal. If AI can do that, we're shaving what would take a expensive human 20 40 60 100 hours plus down into something that can be generated with a model in less in hypothetically less than 30 seconds. So that's another example. Let's pick a third one. So we have outbound cold calls, automated project scoping. The third one could be a uh a rag agent. Now what rag means it stands for retrieval augmented generation. Uh, essentially what it's doing is it's basically just extracting information really really quick quickly. So if I have a massive business um and I've got that I've got all these operations. I've got invoices that I've sent out. I've got bills. I've got people that are moving tasks through project management tools. I've got people sending messages. And if I want let's say for example I'm on a project and the scope there's there's a bit of scope creep and I want to I want to figure out if at any point in time I've communicated with the client. And let's say for example, we said we were going to add a social media calendar into an application we're building. And I think that the start of this project, we said we weren't going to include it because it was going to take too long. But then the client comes back to me and goes, "No, at the start of this project, I definitely remember that we talked about building a social media calendar." So, I go to my rag agent and I say, "Can you please look at every meeting transcript and communication that I've had with a client and with this client and check whether there's ever been any human communication about building a social media calendar?" And what this can do is in in literally less than a second, it can search all of your previous communication history, if you have it set up correctly, to locate that kind of information. So, that's just one example. So, these are three typical uh AI automations that can have serious ROI for a business. I'm talking, you know, if these are set up correctly, this could significantly increase efficiency uh and free up employees to to basically be able to focus on more ROI generating activities.

Now, that's just three off the top of my head. So, how the hell do what how do we uh how do we go about building these with these existing tools? Now, with a context problem, let's take these examples for an outbound cold call, right? Let's say I'm selling uh let's say I'm selling software. So I'm selling application building and I want to train an outbound cold call agent. So it's one thing for me to to go to chat GPT make a custom GPT that goes you are an outbound sales call agent and you are selling customdeveloped AI powered CRM to businesses in Australia for example that's great but then what happens is what happens when it makes a call and that person asks back and goes okay where is your business located and well if I haven't trained the AI agent it's going to start giving away false information so I need to make sure that it's trained properly but then what if it asks okay uh how many employees do you Well, if it doesn't have access to my business content and how many users are in my business, it can't answer that question. And what if it asks what does this process typically look like? How many what what kind of what kind of clients have you done this work for before? Um, what like what are the security protocols and standards in Australia? Um, all this kind of information, right? It needs it needs to be trained on all the information that you have locked in these SAS tools to accurately be able to generate responses.

Now, another example could be a customer service call agent for an e-commerce business, right? So, let's say you have a uh you sell perfume on the internet and in some sort of e-commerce store and you have a customer support agent and people call up and go, "Hey, look, I've I've sent this message. Oh, I bought this product 2 weeks ago. Where is it? It's it hasn't arrived yet." So, what it needs to do is it needs to be able to access your CRM, your inventory management system. Take a look at that specific product. It needs to find the client, find any for find all the products that clients bought and then take a look at the status, right? But it needs access to your tools to be able to do that. You can't just build, you know, a really basic AI automation and get that get that to work because it needs access to all your systems. And that is the context problem. So what happens is businesses if they're in this boat, they are in serious trouble because none of their none of their data is easily organized and cleaned and provided to these models. So these businesses are going to seriously seriously struggle. These businesses, they're a little bit closer, but they still have data spread out across five to 10 different tools. So they need to use make.com in it and Zapia to aggregate all that data, put it into one unified spot and then provide it to these to a a singular spot where these uh AI automations can then go and accurately source that data.

So we have two options here. We have process automation um nadn make.com zapia and the primarily these are often used for now you can use these in both scenarios but this is often used to connect tools together so option one is process automation with existing stack and what we're doing here is we're connecting tools so we're connecting SAS tools now this is not ideal because we're wasting time connecting tools together that ideally should all be in on spot and we can just remove this step entirely. But the reality is is that 95% of small and medium business enterprises are in one of these two categories. They're in this category, they're in serious trouble, but a lot of them are in this category, in which case they're they're all shoot in 2025, they were all shooting down something like this. And so people were trying to build these kind of things in 2025, but nobody had the context and the data organized well enough. So that's why we saw so many failed cases in 2025.

Option two. Now option two. Now option two is a custom application. Now what this is people call this the vibe model. So this is the SAS model. Uh, we have existing SAS tools very rigid uh that all these businesses are set up. The option two is the vibe model. Now the vibe model is that we basically for these businesses we scope out exactly what their business processes look like. How do they acquire CL acquire clients? How do they qualify leads? How do they onboard leads? How do they manage project fulfillment? All those steps, we map that whole thing out and we build them a custom application. And so what happens here is we no longer have to connect different tools together because it's one simple application that does everything that we talked about up here. So it does it's a CRM. It has file storage built into it. It has document signing built into it. It has project management and communications built into it, etc., etc. It's all built in one spot. So all of the data is unified. So there's no need to connect the data together. So straight away we're in a better position where we can then go ahead and provide this unified data back into the knowledge bases that are that are connected to these AI automations. And that's what's going to allow you to create accurate AI agents and automations that are actually able to to source your company data correctly and build agents for these that actually work. And that that's this is the that's the crux or the real problem that emerged out of 2025 which is where everybody's lagging and the companies that can solve this problem in 2026 by either building templates and really good processes to can unify the data using some of these tools and then providing it to an AI brain or going down this approach which is building custom applications which is what we're shooting down which has a number of benefits as well. Um, but the end the end goal out of both of these the major goal is unifying data to AI brain which is often a vector database. This is the main goal that nobody was doing. So the the goal of the goal of these of these two processes and the reason why businesses struggled in 2025 was nobody was able to unify clean the data properly and provide it to these AI models.

So what needs to happen in 2026 and where the real opportunity is is for companies that can find a way to either just work with a couple of specific tools or build custom tools that essentially unify the data. So make the process whatever the business process is very very simple. What if whether that's managing ad ads, whether that's the lead qualification steps, whether that's onboarding, whatever it is, make that process really simple. access the data, clean the data in a way that's easy to understand by these AI models, update them into a vector database or what I call like the AI brain, and then connect that brain to your AI automations, which can be built in Zappy, retail, nenmake.com, all these tools that exist to build, you know, C uh calling agents or voice agents um or whatever agents you're building. And that allows you to provide unified data um provided. to AI brain and connected to automations. And then what this does is it allows for accuracy greater than let's say 98% which is which is realistically the level of accuracy at least you know that we need for businesses to actually adopt this um with with proper guardrails and this is what leads to actual ROI driving AI automations. Um, so this is that process.

Now, how to this is the real crux of the problem that's occurring in 2026 and the real opportunity and how do you actually make money from this? Um, now I've touched on this in a couple of other videos. I'm going to keep this real simple here, but essentially, you know, if you're someone that's building, how the hell do you make money here? Well, it's actually very, very simple. You calculate ROI, demonstrate proof, and reduce risk. So what the hell do I mean by this here? So th this is all well and good. This is the theory around the opportunity that's coming in 2026. The problem that exists. You know these are some examples of you know what's what people are trying to build. Very very popular examples and why they can't build them or why in 2025 they tried to build them and to be honest they were just kind of crap. These are options for how you can unify the data to essentially make these actually workable. And so how do you sell these and why do businesses why do businesses benefit? Well, if we just take these examples which I already talked about before, if you have a sales agent that runs on a model that costs a couple of dollars an hour um and it's its close rate is is ridiculously good. It has access to all your information. It can access, you know, let's take a customer service agent. It can access all that information uh quicker than a human can and it operates at a tenth of the cost. How do we sell that to a business? Well, it's simple. We take a look at their existing processes and we calculate the ROI for them. So if they've got 10 sales agent or 10 customer service people currently um those customer service people cost $40 an hour and that they're $40 a week. Well then what does that look like in terms of gross weekly cost and annual cost? That's how much they're currently spending. Then we go ahead and take a look at what it would cost them if they used this new system. And you know look the savings at that from service level look like they probably be about an 80% saving. Let's assume that they're spending $100,000 a year on human staff for customer service. It's probably a hell of a lot more than that. So an 80% drop is $80,000 a year. That's that's their annual sa savings from that kind of AI automation. Then we demonstrate proof. What do we do by that? Well, let's let's see if we can't build some example customer service agents um and be able to quickly plug it into their business and in seconds demonstrate what kind of functionality that they they could be looking for so they can get an idea of exactly how this technology works. And then we reduce risk by saying, "Hey, look, if this isn't doing what we're intending it to do for your business in 90 days or less, we give you all your money back." Because at the end of the day, if you're selling a quality product that's really doing all this, and it's genuinely saving them $80,000 a year, well, in 90 days, they pro pretty much made their money back. Uh, and if it's really that good, they're not going to take their money back if you're saving them $80,000 a year. So if you're building a quality product, you can go ahead and offer these kind of uh reduced risks because at the end of the day, you know, you're not trying to sell crap. You're trying to sell something that's genuinely saving saving businesses a lot of time and money. And that's exactly how you go about selling this entire thing.

Now, that's what we do at APG Software. We essentially identified this over me doing over 250 projects um both as a freelancer and as an agency in this exact space, building all kinds of apps for businesses. we were doing a lot of this kind of work um but also a little bit of this and because of what's happened with AI coding we've we've naturally come to the conclusion on building the offer that we've built and anyone out there that's trying to get involved start small start with you know start with very very small building converting landing pages using AI coding to build basic applications for businesses it doesn't need to automate all their existing processes if you're just getting started find a really really you know some small projects on Upwork. Now, if you want to get started on Upwork, take a look at some of my other videos or feel free to jump into school and see how we get you on the road in 30 days or less. But essentially, start small, build your skills there. Don't try and chew off any massive complex problems at the very start. You know, get some practice in on some smaller projects naturally, which will look a little bit, you know, these kind of projects. as you develop your skills, you can start to explore possibly building things, bu building your own custom applications um and selling down this approach.

So, this is really the no-brainer way that software development is going for small to medium businesses in 2026. And anyone a lot of people are already waking up and understanding this process and pivoting their software development agencies to address this massive problem because to be honest, pretty much every small to medium business is facing [snorts] this issue in 2026. And if you want to get involved, uh, I hope you find this helpful and let me know in the comments exactly what your plan is to take advantage of this massive shift that's coming in 2026 and exactly how you think you'll do it. Uh, and I'm very interested to hear what other people think about this and how they're looking to uh, address this problem uh, in their own AI agencies or AI freelancer careers. Thank you very much for watching and I'll see you soon.