The Hidden Voice AI Gold Rush (Start Before It’s Too Late)

kind of goes without saying these days that voice AI is one of the best opportunities or areas to get into if you're looking to create value and make money and build a business with AI. So, with 2026 just around the corner, I thought I'd bring in my go-to voice AI experts to break down what they are seeing in the market as being the top three voice AI solutions to make money with in 2026 and beyond. And these are good friends of mine and ex-members of my accelerator program who have really just gone from strength to strength in the voice AI space. So, if you want to listen to it and get it from the source, these are the guys to listen to. I hope you enjoy. — Right, boys. Good to see you. Um, great to have you on the channel properly. Um, if you guys don't know, this is Jiannis and Henrik. Um, both really good friends of mine and absolutely crushing in the voice space, really pioneers of it. So, um, happy to have them come on and really break down what's working for them. Um, what they're seeing for these students as well. So, I'm extremely excited to hear about this. Um, I've got a service level understanding. Uh, there's nowhere near as deep as you guys on these uh, at this point. So, if you

guys want to give us a quick little bit of background on um sort of voice AI space up to this point, I think that would be a good little intro to people who are maybe not as uh as familiar with how the space has gone over the last say 12 to 18 months. — Generally, I would say it's way broader than we all think. I mean, we initially when I started out, I was also like voice is this small little thing, you know, where you call someone on the phone, — but it's so much more than that. We don't just have telephone nowadays. We have like auto narration, we have AI transcription services, stuff like Whisper, stuff that we develop developing on. So it it's just getting bigger and uh the development is insane. Like I got to say and it's switching all the time. What I'm seeing right now is that I get at least one or two inquiries a week of new SAS tools in the voice space. That's actually SAS tools they want to partner up, right? So if you think about back then when I was just sitting there trying to figure something out with platforms like mapping to build a basic voice agent uh you have to figure out everything yourself and now you just go out there and you can create them with AI by yourself. It made a massive difference. — So are you saying they're coming out with like fully specialized platforms around a certain use case or is it like I remember there was this little phase of people's doing SAS around like a tool for voice agents like I think there's one for payment taking payments over the phone. And I was like, that's a smart idea of like picking a specific thing that voice AI developers and people who are integrating these into businesses are going to need, but they don't want to have to do like I want a flawless system that allows me to take payments over the phone. Uh someone else do it and I'll plug it and I'll pay you pay your cost. Is that what you're seeing within the tool layer or is it more so like outright platforms? — Utility things are awesome and there are few that actually do well. I would say the majority is especially people that try to make things simpler. I would say we are still in a point where building really good voice AI is hard. Like building voice is incredibly simple. can make an agent in five minutes. But making it good is the hard part. Everyone just tries to take away this complexity by providing a let's say white label solution that you can install on your own system. You literally just drop in some prompts. You maybe use AI for that and then you have this whole system that you can sell to your clients. Problem what I see and that I think is a big issue and this is kind of like wild west out there right now. It's very general all say like I have the best appointment setting agent or the best lead qualification agent but there's no niche behind it. So they're still trying to find their own ICP. And I think the stage where we're currently at, — yeah, we found people contacting us to get help embedding voice assistants in their own current like SAS platforms. So for example, like a texting or SMS SAS recently uh they want uh people just to click three buttons on their website where they just create a voice agent and then instead of just SMSing, it's now voice and SMS where there is like in the terms of like audit/ consulting forum instead of actually building it out for them because they already have a team. I know the difficulty has often been with the integration into CRM and things like that industry by industry. Do you think there's a space for people to target these major CRM and major platforms within an industry and say like we've got a voice agent that hooks into that? Is that the play now? Or is there something else that's kind of like obvious? — I've seen a few like specific ones um costing about $600 a month just for an integration into Service Titan. If you haven't heard of it, it's like for plumbers or service type industries. If you work out like a connection to a CRM, templatize it. That's uh and then just copy that thing over and over. — Yeah. — Thing you can use it as like a lead machine, right? If you think about you build one integration, it's incredibly hard. You just reach out to them and say, "Hey, look, I can enhance your service like crazy. They literally just throw leads after you. " — It's once you've got that like little hook or that edge to get in on the like get the interest in it. It's like you can speak directly to their problems because once you've like solved that for a few people, you know, like are you sick of like kind of the clunkiness of this or like how you can't achieve these certain things, you can speak so directly to it, which is really working well for people on the outreach side right now. But we can get into a bit of a broader discussion on the voice AI opportunities and stuff. But I want to jump into um seeing what you guys are considering the best voice AI use cases right now that people can jump in and start to to make money with. But I think instead of people trying to be very general about it, we're getting to this point with the AI space right now where there's proven offers there's with millions and millions of businesses that are still yet to get their hands on them that people really if you're starting out with the news or the new stuff that's coming out stick with these kind of use cases that are validated and you know that people have already delivered many times over. So I'm excited to get into it. — All righty. Cool. So the whole goal with this document was to make a very easy way on how we can explain you and everyone watching more on how those offers are actually structured and obviously the best insights. A I think one of the most important things is the ROI. Everything that we do and everything that we teach our students is always focused towards the outcome. You always want to make sure you provide value. So that's the first thing we're going to talk about. Then we just have some general breakdown. The use cases and the agency revenue. The agency revenue is in that case what we actually earned with this. And there are few examples that are pretty surprising. You'll see that real results and timeline. We keep them probably a bit shorter. Same like text stack. We have kilos and client acquisition at the very end and a little bony that is after that which isn't mentioned there but you'll be very excited of that. So I recommend staying. All righty then let's head into

the first one. I think it's one of my favorite ones and I think one of the biggest ones that we've done so far. um that is actually for a fortune 200 company that we worked with and what we did is one of the offers that kind of established itself like throughout a lot of different industries which is lead reactivation. So uh anyone who doesn't know what lead reactivation is simply imagine a business has obviously had leads in the past or even people that probably didn't convert so they just lay around and no one really uses them and we basically help those people to reactivate them. It's kind of something new you see later down the line as well. The cool thing though is the ROI here. That's probably like insanely crazy for this company. probably not as much, but for everyone else looking at it, I would probably say if we calculated very passively was over half a million dollars, $754,000 with everything else we did after that. So that's literally what the company made an extra profit from literally just using that solution that we built in there, which is massive. — So in a sec, I'm going to go through into the use case so it makes more sense. So you actually get an example. — Uh but in this case, we're doing warm outbound calls and the idea is that we they have like a list of people and we basically grab that list and just call those people up. So, is that pure uh outbound voice or have you like slot in some SMS their retries and stuff like that? — Yeah, it was um one call. If they don't pick up, we do a second call and then uh if they don't pick up on the second call, we leave you voicemail message and then we do an SMS and at some always they have the ability to call back. — Sick. — Yeah. So, it's quite thorough. That was a very plain explanation, but uh yeah, figuring that out took a while. Yeah. — Yeah. Yeah, cuz one of the things we've run into with voice AI is the um the timing, right? You want to make sure you like depending on what time zone they're in, you don't want to be calling them in the middle of the night and then retrying at like 6:00 in the morning. — And that that's like the very basic things, but we actually had to take that part that we had to figure out on which days the most people pick up at which time of the day. You know, some are working during the week and they have some lunch break so you can call them in between. You get better ROI on that. It's really crazy. So, you can really optimize a lot of — stuff no one thinks about when you come into voice AI. It's like I mean I guess that's what people are selling on these different SAS tools that are coming out. It's like oh you just sort of slap it together and then a works but there's so much more detail to it. So yeah. — Yeah. So the use case here was we are calling back old applicants that applied to this big company. Um so we were targeting this in this space the talent acquisition niche. Um and yeah the pain points was that applicants aren't hearing back after handing in a regime. So imagine that 8,000 people apply to this company every single month. They don't hear back at all. So our offer was that we help them achieve a task that was impossible by any means before without having like a huge team before without AI. They wouldn't have been able to do it or they never would have been contacted. And so if you've applied for it, the calling back is if they've not been successful or if they have been successful or have you got like some sort of AI shortlister in there that's integrated too. — Yeah. It's the people that were not successful. Yeah. But the idea is that we're asking them are you did you want to continue looking for a job? Did you uh have any other experience that you've gained from the past like month or two? And — that's keeping them in their database. — Yeah. they can like match him with other opportunities within the company. — That was one of our not first project. So we already had quite a bit more of a payment in that case. So that thing basically brought us 54k in revenue. So this project alone — and what was the structure of that deal was like a performance based one or was that just outright build? — Not at all because — I wish it was performance now. — Yeah. Now we have the data and I think that is the biggest differentiator to everything that we do now compared to the initial ones. Like with anything if you don't have the data you don't know how to price it, right? So what we did back then is a handover solution to so we priced them a fixed price and then they came back with adjustments um additional things that we wanted to build out run separate campaigns etc. — And they probably even the thing about that is that they got such a good deal on that 54,000 but at the time they probably was still like you know like I'm going to narrow on it a bit and pushing back like oh trying to get every little scent out of it. So that's just kind of the state of uh AI development up front like that unfortunately. What I got right here, the data we have now, you just showed it to them. You show the numbers and they have no other reason to argue. — Yeah. What are your thoughts on the the performance-based model for lead reactivation? Do you think that's uh superior if they can? And if so, how would you recommend to sell it? Cuz I think the offer around it is really the interesting part. How much revenue should you be asking for if it's reactivating their uh their CRM? Um the stale leads in there, is it like, hey, I'll take 50% of the revenue. But in some companies, like they can't you can't take 50%. this is just not profitable for them. Um, so what have you found in terms of offers there? — You said it right, like it's very hard making a standardized offer because it's very variable. It depends on the industry. Like with these companies here, I know for a fact that they make at least 10 15k on profit on a single referral. So from that you can just take a percentage, right? I would even say with 10% you're really good and you're already doing well. I mean given that you can activate so many of them. for someone smaller or a plumber or I don't know whoever has older leads, I would always make it based on how much a lead is worth to them. So if they can close someone for 3K, you still can take maybe like $300 per lead, which is still like really good if you can close multiple of them. So — you need the data to figure out conservative, but like I imagine like this one would be like 20 30% easy. the results is that in this specific bill we did for like the first pilot or like the first campaign rather we did 36,000 calls and imagine that for being it would have taken 500 full days like 9hour days no weekends no sleep no Christmas uh and an AI did it in roughly like it's a bit hard to say cuz it stopped some days but it was roughly like say 30 days I'd say but I reckon we could have cranked those out in 14 easy — it' be very interesting if you guys actually pushing that amount of calls out to run issues. So, if you could sort of let us know what comes with that amount of scale with voice AI. — Sure. Um, definitely. There's definitely a lot that goes into that and it can break on many different parts. I think one of the biggest ones is you probably burning your numbers cuz if you make too many calls, they often go through a spam and then you can very easily get your numbers blocked and that's something you don't want. So, you need to round robin through certain numbers. And that's also like money. How much uh money that they are willing to spend a day cuz like you know if you want to do more within the short amount of time you got to have more up front cash. — So how many calls a day are they banging there if you spent like 50 days? Um what sort of uh usage costs cuz these very short brief calls or are they kind of dragging them out a bit more? Is it like short brief call to an SMS? Because I think that's sort of where the the goldie like the sweet spot for voice is right. So the cost to run the project was around $4 and a half thousand dollars uh for that specific uh like those 36,000 calls. Um so it was 36,000 calls because we had 20,000 people to call. So it's roughly like one and a half calls per person roughly. If they got through and they finished the screening it was actually about like a 5 to 6 minute call. 32% of people finish the full thing you know like at the end. — Wow. Yeah. I was going to ask about that like the monitoring of like I suppose you need to benchmark it against humans of like how many people are like hanging up or giving up on the call. Um but like the realizing that it is AI and then hanging up is also kind of like the the like ghost that's hanging around these kind of systems, right? So if you're getting them 32% of people is that 32% of people who picked up or all calls got to the end? — 32% of people got finished the screening process. So, they finished those five questions they had to answer and 32% of them finished that screening process. — Yeah. If someone signed up for wanting a job, right, and they are still looking for it, they're most likely keen to still get one. So, it's not like they're calling it cold. And that's the cool thing about it because they're already warm. They know the company and they know they can get something out of that it actually benefits them. So, I think that's one of the reasons that actually led us to have a 30 plus% uh reactivation rate. Can I just mention one cool thing, Lyn, because I think this is the coolest thing about the whole build I'm excited about is that people actually got jobs from this. It's like 58 is creating jobs, bro. Is that what you're telling me? — Yeah, bro. People aren't losing jobs to AR. People are getting jobs from AI. — Yeah, — man. You guys are like the Robin Hood of the AI space at this point, bro. You're flipping that on its head. No, that's awesome.

— Yeah. Lead qualification. It's I'd say the most common use case that we see these days. uh it's you simply imagine someone running ads and you have a Facebook lead form for example someone signs up shows interest speed to lead is incredibly important so if you call them within the first 5 minutes you have a I think 300% increase of likelihood of them closing them as a client or getting them as a lead because if you think about any service that you want to get access to let's say you reach out to someone you obviously just take the first one that comes along right you think they're like the most responsive they're the best and you just ignore the others so you basically by just being slow you basically pass your leads along to someone else which is bad. So lead qualification is super powerful and what I just said is obviously something that bothers people and that's something they can use for marketing as well. So the ROI in this is pretty cool. Uh we basically saved them some money. So this is the side that's not as important as the one that's below that which is how many how much money we actually made him and that's 70k a month from this lead qualification solution. And there's a specific example here that I'm going to let Hrik talk more about. So the uh general breakdown is that this was built for as an inbound receptionist and it just qualifies as potential customers or helps existing ones and it was actually for um it was like a sales rep that acted on the behalf of a real estate company. So people called in and they either like asked whether a property was available and if it was they were really interested it would actually book them a viewing. And another thing it did was a cool little touch is that it grabbed the information of the home like the person that was selling the home and then send the details to the buyer and it also searched for properties. And if you're wondering, we actually use Quadrant for that lean. — Interesting. — Yeah. — So you've got a sales rep who works for a real estate company and people instead of calling him or her and expecting him to pick up the phone, the receptionist is picking up — cuz it's an inbound. So people usually call up and the first sentence they say is like oh is the property available on 123 Main Street and then the AI sales rep would then go let me have a look and then you know either the person then would hang up or book a viewing or mess them with the in — and I've I haven't put a proper like uh vector database and rag system onto a uh a voice agent like that before. What was the how difficult was that to get working well? Um because I know that like Vic databases on their own is like enough of a pain in the ass um at times. So how'd that go? — I I'd say with like everything it's uh it was very easy to set up the initial version but making good it's actually yeah like if it comes to the amount of features that the client wanted for example then you can filter it by the price that number of bedrooms — you have to think about. Yeah. That works actually with retrieval with rack and it's it's a big pain to set that up. Yeah. But um — okay, so you were like putting in like metadata filters and stuff on — metadata filters. Yeah. — On the on extracting the price out, but you have to like make sure you've got the right price and the right and I mean there's all this stuff around voice around like making sure you've got the right numbers or words like confirming things with them. So uh we won't get into that, but I'm sure there's enough um in order to do a proper search, you need to have the proper input data, right? So I guess you had like a couple loops to make sure you get the right input data — and the worst thing is probably it doesn't represent the price. um 5k for the offer of that it would probably if we would rebuild this thing now probably more like 30 35k. — Yeah. — But again this is literally just a result of us starting out. — Yeah. This is I think I get so many you guys probably do as well so many questions around pricing. Um it's always like I how do I price my services? Like I don't know if I'm charging too much or too little. And I mean it honestly depends what route you're coming into this business model with. If you're coming in as say like a a younger person who's just put all this time into learning voice AI and they know it from the ground up themselves and they're basically freelancing then your price doesn't really freaking matter does it like if you're doing it yourself pick an hour rate for yourself and then do it based off that it's no it's not like you're going to go into the red on it but as soon as you if you're coming in maybe as someone who doesn't know this the technical side themselves and doesn't want to learn that then you've got a bit of a tricky position where they might have a dev under them and they're trying to like oversee the build and that's when the pricing becomes important and you just want to make sure you're is breaking even. — Yeah. The problem is I think everyone tries to build a complete solution. That's the biggest issue we see. So what we did is to all of our students, we basically gave them an icebreaker offer. That's how we call it, right? So you literally try to bundle the minimum thing that you can build that still brings value and that's what you push out because you can easily templatize it and you can sell it for the always same rate. So you the main thing is in the beginning people don't have leads, right? So how do they get over it? By offering something that brings value, it's cheap, is easy to implement so that they can get their foot in the door with a company. And I think that's a big learning that we had as well. I would never build something like this as an entry project. It wouldn't just make sense. — Are you framing those icebreaker offers as like okay let's do kind of a validation of this works. Let's do a PC MVP of this and let's see if you're happy with it or is it still like this will still be plugged in because it's usually that integration step and putting it into production. So are you selling them like a one two threek MVP just to like get their feedback on it and validate that it works and with a basic set of like uh quering abilities in the database or is it a sort of different offer entirely? — Yeah, for like anyone watching I'd recommend not hooking up to difficult CRM and anything like that. Um so separately on air table offer something like um we're going to reduce the amount of spam calls you receive by 80%. And like the only job it does is like if it if it's uh if it's a spam call, we just hangs up. — Yeah. Or take out of office hour calls. Like anything that's simple that's really simple to implement. — Okay. So, you're still going to swap out say like that point of entry to uh like replacing their front of house number with your one, but it's just going to cut the spam calls and then transfer to human on top of that. — If it's a qualified leak. Yeah. — Yeah. They all need they think they need a big system, but they don't. Like in the beginning, it's just getting exposed, gathering data. Okay, data is the most important thing like I say without with data we wouldn't have charged a $5k only for this solution that's literally just because we had no idea we had to trial things right so that was definitely worth it — but I mean you still have to go like people still got to go through that you know like — well the thing is you have mentors right if I would have had someone like me you know that with that experience now it's incredibly easy to just build an offer because you can tell them what to do like stay away from patient sensitive data that cost you so much time you have like after hour reception is that you can get out there. There's tons of stuff you can do nowadays. So that this stuff doesn't happen. I mean, it's it works too, but it's painful. It's really painful. You can tell you can ask Henry. He's he wasn't happy for a reason. Hey. — Uh, so for this one, it on average like 2,000 calls a month. Um, 32% of the calls were and 6% of the calls altogether, 38% of the calls that were like quarter quartile leads. The rest were transferred or the rest were other. um 38% of that 2,000 is 760 people. And for this like use case, we're going to assume the lowest possible close rate we can think of. So like when they get transferred to a human, um that person will then actually sell that home and then their real estate agency will then actually on average get about 10K like 2 and a half% of like $400,000 is about 10K. So 7* 10K will give you about 770,000. This is where we got those numbers off originally. this 1% — way more but we're just being conservative right — the other things were like people just you know calling up hey is this house available and literally after the golf I just hang up um — people also just call up and they just like don't say anything super weird and then the other one was people just selling like was Google ads most of the time it was that um and that took us about two months from that time of on boarding table solution work in the wild was pretty hectic two months there

— yeah customer success uh it's a very broad term there there's a lot that goes under that for that example we've just chosen customer success cuz there's a couple of insights that we want to give you along that are relevant for a couple of use cases that you can build out under that umbrella and uh you can see based on the ROI that this is more focused towards savings so not on really making money which makes it a bit harder to market but the people who struggle with that they definitely see the value in that so saving some company 200k a year is amazing uh if you know how to prove it to them right so the numbers here are I would say very special bas based on the client that we got uh to actually make this happen. So these are actual real numbers. We actually save them like around 200k a year and we given them 8,400 minutes back uh per month and the way the works I think is broken down anyway. So Henry will probably share some more details on that. The general breakdown of like a customer success is cuz it's an inbound setup and I love inbound setups way more than outbound stuff. It's a lot quicker to set up by the way guys if you're wondering building out. — Is that cuz you don't have all the like scheduling things we talked about before. uh outbound just regulation and yeah like the call cadence is really difficult. So if you call someone once, how do you figure out whether it was a voicemail, robo machine, whether they actually picked up or hung up? Like there's so many little variables. This setup involves helping out a lot heaps of things. So the prompt usually is massive and a single prompt, by the way. um you're answering common questions, you block spam calls, you transfer uh you gather info, you update automatically people's CRM, and you send updates to a live team, whether like through Slack or SNS or whatever. And in this case, the use case was um we built it for a high ticket ecom store. They sell things for like between 30 and 50K a pop. They wanted to upgrade an inbound IVR system from a dinosaur to like a futuristic one, which is what we helped them with. — Yeah. And like dinosaur, you got to say like Henrik built insane scenarios for workflow automations what they had as their IVR automation. You can imagine it's similar to a workflow. It's crazy. It's the Yeah, I built spaceships for like my scenarios, but this thing was like an inception of like 200 of them. This thing was nuts. And we replaced it with one prompt. Um so this is where actually our numbers came from like the 200k in Australia. Um you know, an average pay is like 440 to 50k. Um, so they had five to six employees sitting in different departments was like billing uh sales parts or whatever it was and now they just have one customer lady taking some of the edge cases. — They had like a typical ecom like email support that people could email and then it would like kick off a bunch of scenarios based off that or what? Or is it actually like a phone based one? Oh, cuz it's high ticket. They must have had phone, right? — Completely uh phone based. Yeah. — Okay. — Yeah. There's some more advanced stuff in there that Henry didn't talk about, but you could, for example, locate certain places or areas or offices via a zip code. They also have like 4,000 documents, huge PDFs, and like the domains like four and a half thousand links. So, that all works nicely with the voice and the rag system. — So, is that just default rag on something like Vappy or retail or have you got a custom one set up? — Always custom. Never use Vappy or the retail ones. the in-house solutions pretty much puts in like two pages worth of info per turn into the call and with — is that even when you're not calling it? — Yeah. It would just automatically — just default context fluff by turning the knowledge base on. Yeah. — Yeah. It's not good. — Yeah. And like over a 5m minute call what happens to the voice when it starts to reach like the context limit it starts to go like metallic and really robotic and seems it sounds drunk almost. Um yeah. So we use just like a custom tool call to grab in very specific like two sentence max info and bring it back. — It also depends a little bit on the client, right? Like generally if you have a bigger client, they also want to have everything hosted on their own text. So they might already run on Azure or an AWS. So you literally just use their existing server infrastructure to implement those rack tools right in there. This agency revenue pretty much clear. I think this was also one of the bigger builds that we had. Uh, this as well is basically a result of a an entry offer which was probably around 26K and then tons of follow-up offers with adjustments. And I think if we head down to the results, you'll also see that it's kind of like a never- ending story, which is something that we teach as well. But um, yeah, I'm going to give this part to Henry cuz it's going to be super exciting. So, they get like two and a half, two, like in this month specifically 2,800 calls. And its main job, this specific system is to just transfer or route people. So, it does 2,000 transfers, not with 99% accuracy. Um, with some cases like Twilio failing, there was like maybe two or three of them. Um, these are the actual numbers. So, if you guys want to see how much it actually costs a business to run this, um, — $324, absolutely nothing for a whole month, like for that many calls. It's it's like what would you say? — So, that's cut down from like six people down to one person plus 300 a month. That's crazy, man. — It's basically something that is that just guides us on our journey pretty much, right? Um, we still we're still in the trenches of having so many leads that it's very hard for usually to take on new ones. So, we are very selective. Uh, which is also a good thing because it just tells us, you know, there's so much demand in the space and it's just like growing and growing. But, uh, yeah, you can see from a client like this alone, we could probably have like a full-time team literally just working on that because they come up with more things and more things. And that's specifically what I referred to in the beginning. If you talk about a handover solution, which was actually what we offered there as well. They fixed just one time rate. they come over with more and more things and it's basically like some sort of MR setup just say we've got the exact same thing at Morningside. We don't necessarily have clients on retainers but we don't need to. We have them on essentially like they're on like self prescribed retainers where they're like okay here's take another 20 30 grand every month because we want you to extend the system. So that like on the sales side takes out so much fluff and from like the contracting and trust building. If you can get a work your way up to getting big clients that are solid and they have a lot of needs and you have a team capable of fulfilling on like being flexible enough to be able to not just deliver one template but also kind of go outside of that. It's the dream cuz you just go oh yep well like hey we either you suggest it and say we reckon you could add this on or they come and say hey could you do this and you know you've really like crushed it when they're coming to you and saying hey but like could you also do this as well and that's really if you have like five clients who are just doing that and they roll over each month it's the best spot to be in my opinion. Yeah. And it's such a treat because it's literally just for doing a great job in what you anyways do, right? — Yeah. And you get to build a great relationship with the client as well. You've like made a meaningful impact in what in their business. Um and it's just great to work with great people. — Tic tac.

— Yeah. That's obviously like a very crazy simplification but I think that that's what suits this channel because most people already struggle with literally I think even for the beginners this is already too much cuz uh I can tell you from experience that people can't decide on taking maker or in the same thing careless for retail or anything else around it but — like months trying to figure out which one to use — month it's actually month yeah it's like not even exaggerated like there's people that if they spend months they usually just drop off because they can't stay consistent with having a single decision it's insane but yeah this is basically the overall thing that you need to just build any kind of solution out there at least as a PC or MVP. So minimum viable products. This can really get you there. — Just a voice AI platform, a tools platform and a database. — Yeah, pretty much. That's it. — I always get this question which is how do I deal with if you have say a client who you get in touch with or a potential client and they're saying, "Oh, well I went and looked into their CRM that they're using and there's no like ways to hook into it. There's no APIs available to it. What is your answer to that question? I know we've talked before and you're like if you can just build it kind of siloed and separate where it's built into your air table and it kind of runs uh separate and then you maybe charge them like an extra maybe 20 30 40 50k for either integration into their CRM or to move them onto another one that's more AI friendly. Yeah, it's a very tricky one. Uh there's three main ways and I think u a the hardest one of all of them is probably building out an automation that can be like a web browser based automation that builds an API on top of it which is really damn hard and I would probably not even touch any of that like at below 50K like not at all. Uh the second one is that you move them to a different CRM which is also not the easiest because people don't like change especially in a business and usually it's also a lot more expensive so it's also not an option. The thing that we've seen work best for that specific use case is that you set up a system for yourself and you basically just offer the services that you anyways offer through that separate like completely isolated system and you just hand over the final end results. That's usually the thing that works best. So if that comes for example to I don't know like lead qualification you just take a go high level CRM systems you have your whole automations everything nurturing set up through there and when a meeting is being booked you just book them into a calendar of their Google calendar they still can convert them and if they're important enough you can always put them back into the CRM system. — Mhm. Interesting. Yeah. I suppose the other the final option there is to just try to find a type of client that has a CRM that you can integrate with. You know like some people think this is the only person I can sell to on the planet. either like go somewhere else and find someone who's willing to take like a air table based system or um it might be going to them and saying hey look we would have to move on to onto something else here and that's like that's just how it's going to be. So — yeah, it's another — you can just put them in front of the decision, right? That's a cool thing. Like you don't need to completely ignore them right from the start. You can just tell them look it's very hard. You it's it cost you a lot of money. It's probably not worth it. So why not just let me take care of the work and you literally just get the results over to your side and you just pay like a monthly return of XT for it. — And this factors in stuff like company size, right? So the smaller they are, the more likely they are to shift over to something new where you could say, "Hey, look, I'll just like is do you think there's an opportunity for people to kind of build almost a custom CRM that integrates with this sort of stuff? " Because there's a lot of this custom software talk going around now where as an agency, if you're working with these smaller businesses, you can get to the point where you do have kind of a custom stack that you plug in. So like with the stuff that you got currently, you're paying all that money for the CRM. I can build you a custom one here. It's going to work way better. and then you get your hooks and them like in a significant way of like they're you can get them on retainer and they're kind of reliant on you for managing that moving forward. — Probably not like from for at the beginning I'd highly recommend keeping it super simple like — yeah I mean this if you're like maybe 10 projects in and you're saying like okay well I want to be able to wipe this out and maybe have a bit more control over it like in server it's like setting you up on a basically better AI like rails for AI. So it's more of like a an offer where you're offering this kind of transformation to a more AI forward- facing version of the company where you're built on like better rails that agents and automations can hook into. — Yeah. In that case, it would make sense if they really understand the value proposition. So it is you having to build out that solution which would take probably a team of devs for a few months, — right? I think Henry, we talked many times about it like you can have those like some semi custom solutions. It's basically this whole funnel of how you end up with IP which is kind of like our biggest goal always, right? having some sort of something that you own that you can sell off. And uh it starts always with an agency, then you templatize stuff, then you maybe move it to a more customized solution and then you basically just have a team of devs creating small pieces of that as microservices. That's how we call it until you have basically everything transferred into a custom system that you can actually exit. The exit for us is pretty much like the biggest thing that you can have as a an agency owner.

Mhm. — So what worked for us is templatizing the high performing solutions, the outbound lead reactivation one. Built it at once, copy paste pretty much. Our prices were premium and people were just throwing money at our feet like no issues because we understood their pain and then they knew that we understood what they wanted and then uh they felt like they knew what we were going to build out for them and then they trusted us. Because of that we also partnered up with me and Annis. So partnering up as an agency as a solo one as a solo agency owner would be really hard to scale this thing. You need to partner up as — it's got to be like I don't really know any people who aren't doing it with at least a twoman team. like there there's just so much to it — more towards I find consulting and audits we get more people asking us for help and then what didn't work is the cold outbound builds you are relying on the person's uh lead quality list and if that's then the build is then in turn also — um and accepting complex builds initially we actually talked about this just keep it very simple MVP like phase one do something very quick for like max one month build like no longer than that and then we didn't hire enough devs early on before we started scaling. So when we reached the point where we had like 16 projects all at once that was too late and then when we started hiring it was too hard. — And where you guys hiring out of mainly like what's the best places and like how are you vetting that they know what the talk about? — Yeah, that's a very tricky one. — I would say it's a whole video and it's probably — yeah it's way more than that. Yeah, it's — if you're technical and you know how to ideate and you can outsource literally just the work that's pretty much as processed or conceptualized like an SOP as possible, you can find really damn good desk for a very cheap amount of money. The hardest part is basically this ideation process on a you understanding the text needs to go into and how to execute that. — Yeah. So, uh I think that's pretty straightforward, right? We get and we probably seen that through your channel as well and through your community. Uh

clients are freaking easy to get through YouTube if you literally just show off your skill set and you just provide value. It's I think still the best way on getting any kind of clients because you also don't need to deal with any objections. — Voice AI is also still like quite untapped I think on YouTube when you look at like NAM workflows and like all that stuff is just absolutely like chocker with stuff now. Um voice AI is still I mean in my opinion probably better place to get into right now than going into that highly commoditized place. Well, it's definitely getting that way. Um, — there's still not that many voices uh like you guys on voice AI and with on YouTube. — It's very surprising cuz I'm a big believer in it. And you know, for me and I think that's it's generally clear, voice AI is for us the quickest medium of communicating generally with technology. That's not going to go anywhere until we have brain implants. So, it's just accelerating. And yeah, at

the very end there's one thing that I wanted to mention which is a very new kind of offer that we actually see inside of our community quite a lot and it's very interesting. It's something that you actually mentioned on your channel too. I think you called it like the consultation part, right? Where you basically are an AI consultant. — AI transformation partner. — AI transformation. — Yeah. We just do the same with AI audits. It's literally the same thing. Business owners are so damn afraid of not understanding what where to go into like how to start, what to build out, what is actually feasible, how much does this stuff cost. If someone has knowledge about that, it's very easy for them to basically just start building stuff out. Like we have also people in the community like Liam, he's nearly making 20K a month from doing exactly selling these AI audits. It's crazy. So it it's such a good offer and I can just emphasize like if you go into AI audits or any kind of consultation, the doors are open to really templatize it and make a lot of money. — Yeah. The interesting part about that though is um it's a bit of a chicken and the egg issue of like how do you know the stuff before you start teaching the stuff and consulting on the stuff. So what's your like rapid AI consultant uh blueprint or do you think they still need to sort of roll up their sleeves a bit get in there learn the stuff? I think this is a really big question in the industry right now cuz people are going to start pushing the consultant thing and I'm still kind of on the fence about the best way to train someone as a consultant without actually doing the stuff themselves. If you have no idea what you want to do, you just try to build a solution for yourself in a technical way, right? Whatever that is, if it's like building a voice agent, you use Vapi. The main thing that you need to get out of it and I think that is literally the only thing that really matters is understanding the concepts and the structure. Like if you understand like think about Nad for example, you have a workflow, right? If you talk to someone and say this is an automation, great. Yeah. But it's actually a workflow automation has a trigger or multiple triggers that connect to actions. If you understand how those things communicate, you learn so much that you can replicate to any kind of system. And this little piece, this knowledge on how to connect those things is actually what makes the difference of you just building something compared to you literally just opening a new tool and understanding it right out of the box. Like I built these workflow automation tools prior, right? It doesn't matter which new one is going to come out in the future. I will — saying if you maybe can go in there, I always say people look if the bare minimum find tutorials and I try to give people the tutorials. There's so much on YouTube now of follow a tutorial along end to end on each of these use cases we've talked about and at least get like a an initial version set up that you can play around with and test and that should give you enough of an understanding of how these things connect together how you pushing data to air table if you want to set up a rag system that's how you'd set it up how to handle orchestration and like sending calls at different times if you can get a rough grip of that and understand what maybe the top five to 10 use cases are for some like voice AI then you're probably ready to start hopping on calls with people and just put a refund guarantee on Yeah, watch YouTube videos, get a mentor and break things on purpose. It's one of my favorite things. They work incredibly well. — Yeah, I also try and gauge uh where at what stage that person is like in a developmental journey of or understanding voice AI. Like recently, for example, I had a guy on LinkedIn reached out and he's like, I need my setup to be on life kit and with us like how many people actually going to be able to set that up? It's like he wouldn't jump on a call with him, right? But I said 450 an hour. He's like, yeah, sure. And he just threw money. So like um like no one can really do that. — And I will point out that we haven't really touched on yet, but Henrik, I remember you coming into the Excel with like absolutely no like technical background or anything. And so if you guys are feeling like, oh, there's these two like career developers on here talking about voice and how easy it is to get in. And Henrik has like I've seen you bash your head into this automation stuff for like what 18 months to it's really get to this point, but you've done it and it's been freaking incredible to watch, man. And obviously Janna has been a great uh a great mentor for you to for few to work under and join this uh thing that you guys have been doing. Um but it's just goes to show that like anyone can really come in and if they really put their mind to it, learn it. And I'd say I think you'd agree Henrik that it's getting easier and easier now with more information being around, more clear use cases being there um for people to follow the same path that you've gone through. — Yeah, I appreciate that Liam. I mean I think it was all due to you and your program actually. I wouldn't be here without it. So without that definitely wouldn't have been where I am right now. Thank you. — Appreciate it, man. Well, it's been a pleasure to watch, man, and you guys are absolutely crushing it. I think we can wrap it there. Well, that was a an hour that flew by. I think that's an absolute master class for the voice AI side. Always great to chat to you, boys and um looking forward to seeing you soon. And if you guys want to get in touch with Jon Henrik, I'll leave the details down below. Um but it's great to have you guys on. Appreciate your time. — Thanks. X.