The argument in one line.
An AI agent built as a short branching qualification flow — texting like a human, asking five questions or fewer, and suggesting call times instead of links — can book sales calls around the clock for a fraction of an SDR's salary.
Read if. Skip if.
READ IF YOU ARE…
- An agency owner who currently pays a sales development rep to qualify and book calls for clients and wants to cut that cost.
- A business owner generating leads through paid ads or organic content who has no one following up on them fast enough.
- Someone comfortable following along inside an actual flow-builder (branches, triggers, nodes) who wants to see a lead-qualification agent assembled step by step, not just described.
SKIP IF…
- You want a jargon-free explainer — this assumes you're watching a live CRM/automation interface, not reading a summary of concepts.
- You need outbound cold-calling or cold-DM prospecting — this is entirely about qualifying and booking leads who already opted in.
The full version, fast.
The video builds a text-based AI appointment-setting agent from scratch inside a platform called Appointwise, aimed at agencies and business owners who currently pay $3,000+/month for a human sales development rep. The agent runs as a branching conversation flow: an opener confirms identity, a handful of qualifying questions filter out bad-fit leads, and a booking branch suggests specific call times rather than sending a calendar link, which raises conversion. Follow-ups are staged twice within the first 24 hours to catch distracted leads, response delays and split messages make replies feel human, and a knowledge base stops the agent from inventing wrong answers. The system connects to GoHighLevel via a trigger event, and one real deployment converted 36% of 560 leads into booked calls.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 03:10
01 · Appointwise Intro
States the cost problem (hiring SDRs at $3k+/month), introduces Appointwise as the tool, and walks the integrations page (GoHighLevel, Twilio SMS/WhatsApp, Meta, Zapier).
03:10 – 16:44
02 · Building the AI Agent
Builds a full qualification agent with WiseOne from a plain-English brief: intro message, identity-confirmation branch, qualifying questions, a qualified/not-qualified branch, and a booking step that suggests call times instead of sending a calendar link.
16:44 – 30:59
03 · Testing & Fixing the Agent
Runs live test conversations, catches a logic bug where the agent disqualifies leads too early, tunes the objection-handling branch, and reviews personality/tone, response-timing, and dynamic follow-up settings.
30:59 – 33:22
04 · A More Advanced Build
Shows a real client agent with per-service branches (kitchens, bathrooms, flooring, etc.) and its live analytics: 560 total leads, 183 active, 36% win rate.
33:22 – 41:54
05 · GoHighLevel Integration
Connects Appointwise to a GoHighLevel account: imports a settings snapshot, wires the API/client keys, sets a contact-tag trigger, and builds the matching GHL workflow that adds the tag when a Meta lead form is submitted.
41:54 – 44:32
06 · Extra Features
Covers ElevenLabs voice-note integration, full conversation history and human takeover, pipeline management, and the self-learning WiseOne model that improves from every conversation.
44:32 – 45:16
07 · Wrap Up
Closes with the SDR-replacement pitch again and the 10-minute response-time stat, then a subscribe CTA.
Lines worth screenshotting.
- An AI appointment setter can replace a $3,000/month SDR for under $100/month while working 24/7 across SMS, WhatsApp, Instagram, and Facebook Messenger.
- Sending calendar links instead of suggested call times measurably lowers booking conversion because it reads as less human.
- The fewer qualification questions an agent asks, the more leads book a call — every added question raises lead quality at the cost of volume.
- A roughly 30-second delay before an AI agent responds, paired with splitting long replies into multiple messages, makes automated conversations read as human rather than robotic.
- Following up a cold lead twice within the first 24 hours, around 3 hours then 6 hours after no response, catches leads before they cool off or move to a competitor.
- Businesses that contact leads within 10 minutes of opt-in see roughly 21 times higher conversion than slower follow-up.
- A single home-improvement AI agent handling 560 leads converted 36% into booked appointments.
- Splitting one generic qualification flow into per-service branches (kitchens, bathrooms, flooring, etc.) personalizes the questions and lifts conversion at scale.
- An AI agent's knowledge base, built from a scraped website, uploaded docs, or pasted text, is what stops it from hallucinating pricing or service details.
- Writing-style guardrails, like banning em dashes, can be baked directly into an agent's prompt so tone stays consistent across thousands of conversations.
- Routing a disqualified lead into an objection-handling branch that asks about budget or timeline, one at a time rather than both, can requalify and recover otherwise-lost leads.
- When a human takes over a conversation from the AI, the agent needs to stop responding automatically or the lead gets two conflicting replies at once.
The branch script that books calls.
AGENT DESIGN RULES
Booking more sales calls with AI comes down to a tight qualification branch, human-feeling response timing, and a knowledge base that stops the agent from hallucinating.
02Building the AI Agent
- A lead-qualification agent is built as a branching flow: an intro message, an identity-confirmation branch, then a short, required set of qualifying questions.
- Keep qualifying questions to five or fewer — every extra question filters for quality but shrinks how many leads make it to a booked call.
- Suggesting specific times instead of sending a calendar link measurably increases booking conversion because it feels more like talking to a person.
- A knowledge base built from a scraped website, uploaded docs, or pasted text is what keeps the agent from inventing wrong pricing or service details.
03Testing & Fixing the Agent
- Testing an agent by simulating a real lead conversation exposes logic bugs immediately, such as disqualifying someone before they've reached the actual qualification branch.
- A roughly 30-second reply delay, plus breaking long replies into multiple shorter messages, makes an automated conversation read as human rather than scripted.
- Following up a non-responsive lead around 3 hours later and again 6 hours later, within the same first 24 hours, catches them before they cool off or move to a competitor.
- Writing-style guardrails, like banning em dashes, can be baked directly into the agent's prompt so its tone stays consistent across thousands of conversations.
04A More Advanced Build
- At scale, splitting one generic qualifying flow into a branch per service or product line gives each lead uniquely relevant questions instead of generic ones.
- One real home-improvement agent handling 560 leads converted 36% into booked appointments — evidence a tuned branching flow can outperform a generic script.
05GoHighLevel Integration
- Connecting a qualification agent to a CRM is usually just a trigger event, like a new tag or form submission, that fires the agent and syncs the conversation back.
- Reusable account setups can be packaged and copied between CRM accounts as a single importable snapshot, cutting new-client setup from hours to minutes.
06Extra Features
- Adding a voice-generation API key lets a text-based qualification agent also send spoken follow-up messages, which can lift response rates further.
- When a human takes over a conversation from the AI, the agent needs to stop responding automatically, or the lead gets two conflicting replies at once.
Terms worth knowing.
- Appointwise
- A paid AI agent platform used in the video to build text-based lead-qualification and appointment-booking agents that run over SMS, WhatsApp, Instagram, and Facebook Messenger.
- WiseOne
- Appointwise's own large language model, trained on the platform's aggregate lead data, used to auto-generate a working agent from a plain-English description instead of building the flow node by node.
- GoHighLevel (GHL)
- An all-in-one CRM and marketing platform that combines a website builder, email/SMS marketing, pipelines, and appointment booking, used here as the system the AI agent connects to for lead data and calendars.
- SDR (Sales Development Rep)
- A human hire whose job is to call or message leads, qualify them, and book them onto a sales calendar; the role the AI agent in this video is built to replace.
- Smart Branch
- An if/then decision node in the agent builder where an AI prompt, rather than a fixed rule, decides which path a conversation takes next.
- Snapshot
- GoHighLevel's mechanism for copying a full set of account settings, workflows, and automations from one account into another, used to quickly replicate a setup for a new client.
- Dynamic follow-up
- An AI-generated follow-up message written fresh from the specific context of a lead's prior replies, rather than a generic templated nudge.
- Meta lead form
- A native lead-capture form on Facebook or Instagram ads that collects a prospect's contact details without leaving the platform, used here as the trigger source for new leads.
Things they pointed at.
03:20toolMeta lead forms (Facebook/Instagram)
03:45toolTwilio
03:50toolZapier
42:40toolElevenLabs
Lines you could clip.
13:40
“My recommendation to you is try and avoid sending calendar links. If you send calendar links, you're gonna get far less conversions on the back end. So always try and suggest times. It makes it feel a lot more human.”
28:20
“We strike whilst the iron is hot.”
45:00
“They increased their lead conversion rate by 21 times when they reach out to all those leads within ten minutes. Humans can't keep up with that.”
01:45
“It made such an impact on me that I chose to actually invest in this company at the start of last year because I want a slice of the rocket ship that these guys are on right now.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphoranalogystory
00:00Today, I'm gonna teach you the cheapest and easiest way to convert leads into booked appointments using AI in 2026. And this is for you whether you're an agency owner looking to book appointments for your clients and make them more money, or you're a business owner trying to do the same for yourself.
00:15Now quickly, what do I mean by an appointment? I mean, a sales call. We take our leads that we generated from maybe paid ads or organic.
00:22It doesn't matter where they come from. We reach out to them, qualify them, warm them up, and book them straight into the sales calendar where our team is then going to sell a product or a service to those individuals. Now before AI, we would have to hire sales development representatives for $3,000 a month plus in order to spend their entire day calling for leads, setting appointments regardless of what industry we're in.
00:44It's very expensive. Today, I'm gonna teach you a strategy that you can use initially for free, but ongoing, you can reach out to thousands of leads for under $100 a month.
00:52This changed the game for my business, and it's gonna change the game for yours. So this is a system we're gonna be using today. It's called Appointwise.
00:59It's the same myself, and thousands of my clients have been using for over a year now. It made such an impact on me that I chose to actually invest in this company at the start of last year because . So all need to do is go over to apointwise.io and claim build my free AI agent.
01:15We can try it for free for fourteen days and actually get return on investment during that time period before we even spend a single penny on the platform. Now the first thing we're gonna see when we get into the platform is the integrations page, and we can connect straight through to a CRM system, one Go HighLevel, which is a CRM that I use.
01:31I'll actually go through the setup of that later on in this video. In fact, I'm gonna go for a full a to z setup today and also highlight some of the features that I've been using to make the most returns for my clients. We also can connect straight through to SMS and WhatsApp if we don't use HighLevel using Twilio, and we can connect straight to through through to Facebook Messenger and Instagram DMs if we wanna handle all of those inbound messages through a meta integration.
01:54And if you use a completely different CRM, you can also use Zapier or the back end API to integrate custom with your tool. So we're gonna ignore everything that we see on screen right now. We'll come back to it later, but for now, we'll skip straight through to the meet and go to AI agents, hit new agent, we get three choices.
02:10We either start from scratch, we start from a template, or we build with WiseOne, which is a point wise's own large language model that's trained on millions of leads across the platform. Uh, let's have a look at some of these templates to start off with.
02:22So we've got a universal appointment set up. And, uh, here, we can see this is a general flow based system. So if you've ever seen flows on on CRM systems or automation platforms, this will be familiar to you.
02:32This is completely alien. Do not worry. I'm gonna go through everything step by step in this video.
02:37By the end of this video, you're gonna have full understanding of exactly how everything works so can get the most out of the platform and start converting more leads. Okay?
02:45And so we got lead activation. Okay? Reactivating leads that have gone cold in the database.
02:49This, by the way, is a is a great value add to any agency's outlets. Plug this into the existing database, start printing money on old leads. Your clients gonna absolutely love you.
02:57Appointment, no show recovery. We've got niche specific ones as well, so roofing qualifier, front of house, as in front of house, all that we got there are triages and so on for restaurants, hotels.
03:09And so we can go through any of these templates. So what I'm actually gonna do is I'm gonna go ahead and build with WiseOne, and and then we will also have a look at how we can build manually inside of the builder here as well.
03:22Okay? And so over on the left hand side, we're introduced to WiseOne, and we get asked, what do we want this agent to do from the first message through to the final outcome. I want to convert leads who opt into my meta lead form ad and book them on calls.
03:45This is the most common way that we use Appointwise in ourselves. Clients that are generating leads don't have the time to reach out to those leads. It's expensive for agencies to hire SDRs, and so we use Appointwise in order to reach out to those.
03:58What is the business called? Just get a local business to me. K.
04:03Let's use Anglian Home Improvement. This is a genuine business.
04:08First name of the rep, let's say the name is James. What area do we serve? Norfolk, UK.
04:16And these leads come from, yes, meta lead form only. We could do Instagram DMs and so on, but we're gonna create something really simple here.
04:25Nice. James sending the messages serving Norfolk. Next bit, what makes a lead good for you before James tries to book them in?
04:33Who should we rule out early? Must be a homeowner, must have a project budget of at least 10 k.
04:48Let's say this is high ticket and must be interested in doing this ASAP as soon as possible.
04:57Okay. So these are gonna be the free qualification criterias. Regardless of whether you're doing this for a client or for yourself, we wanna make sure that we qualify all of the leads.
05:05All every single lead is created equal. We don't want us to be booking appointments with ship. When someone's a fit, what should happen?
05:11Book them directly. I wanna make sure that you guys can see this on screen. Hopefully, you can see everything nice and clear.
05:16I mean, you might need to go full screen on this video. When someone's a fit, what should happen? Book them directly into the calendar, send them a booking link, hand them off to a person's schedule.
05:23I wanna book them directly into a calendar. Which calendar should these bookings go to?
05:30On default, the call links are twenty minutes.
05:35Don't connect calendar for now. I haven't set up an integration.
05:41You would usually go through and set up an integration first, but I'm not doing that for this. We would skip the calendar integration for now because it needs to actually connect to the calendar, see our availability so it can actually suggest times to the lead before it books in. If the agent gets stuck, real objections, someone asks for a human something of script, shouldn't it hand the conversation to a person or keep trying to handle it?
06:04Try to handle once or twice and then hand off to a human.
06:11Great. Send it to my email address. I'd like the agent to sound warm and consultative.
06:19We could have friendly and upbeat. We could also send our own personalized version of this. This is the tone in which it's gonna converse.
06:28And, no, I don't want anything specific for it to lean into. Any FAQs, pricing sheets, policies we want to cite?
06:37I will add knowledge base later. We'll add the knowledge base later on, and it can scrape the website.
06:47Great. And so what I mean by knowledge base, if you are completely new to agents, is for this agent to be responding in the best way possible and for it to be the highest quality responses, somebody asks questions about Angulan home improvements, I don't know, windows and door fixing service. We need to have information on the back end so the agent doesn't hallucinate and actually states the right information.
07:06So we will go over to knowledge shortly, and and we will go ahead, and we will we will give as much information as we need to to this agent. Yes.
07:15Please build this. And now it's gonna go ahead and build this for us. Okay.
07:20So the workflow is now being generated. Let's have a look at this.
07:24So if we go into this, we can see that the start, the source is meta. And if we scroll down, it's automatically being set up for us. So we've got the name of the node, so we can name these nodes for organization.
07:35We can see it's an external trigger. The source is gonna be meta, and the event is gonna be meta form submission.
07:42Okay? Now we're actually gonna set up go high level, so we will change this trigger. But for now, it's via meta.
07:48But you also see we can set this up by API. So if you're if you're used to using Claude code and systems like that, you can just trigger straight from the API or Zapier, for example, if you're using a different CRM system. The event is gonna be meta form submission, and so that is what the trigger of this is.
08:03We're then gonna hit an introduction. So, hey. It's James from Angulan Home.
08:07You'd asked about our service. Just making sure I've got the right person is this first name. Okay?
08:13And this is a really nice simple opener. Another opener that I would recommend to you is if you have a very easy binary question. So you could go straight in and be like, just wanna check your homeowner right now, question mark, and you want it to be a very, very simple.
08:27The first question wants to be a yes, no answer if you want the highest response rate regardless of what you sell. So if that's a question, we're gonna keep it how it is for now. And we've got AI instructions.
08:37Send the open. This is a scripted identity check. I'm gonna save the answer intro response.
08:43If you wanna get advanced, you can change these variables, and it's gonna save variables on the back end of your CRM system. So you're saving that information. You're basically enriching leads.
08:53So when the team actually go to call them, you've called all of that information in there, and you can set how many times you want it to retry when it's a required step. And sometimes it will it will skip that step if you don't have that set. How we want the the message how strict we want the message to be.
09:08And so that basically means, can we allow the AI to go off script? We also do have our merge fields in, so we can personalize this message by using any of these merge fields as well based on information that we've got from the lead. If we get pushed back and they say, why?
09:25Who is this? Then we can set what we want to happen. Do we wanna acknowledge and reask?
09:29Do we wanna explain why, or do we wanna skip on the pushback? Acknowledge and reask would be the standard.
09:35And in fact, for most of you guys watching this who have no experience building AI agents, I would keep the default settings that we have on right now, but I'm gonna make the efforts to try and explain everything to you so you've got a good understanding if you if you do want to go ahead and play with it in the future. Then we have a branch.
09:52Okay? And so a branch is really an if this, then that logic. Okay?
09:56And this branch is, do we have the right person? Because the first question is, is this, you know, Jordan? Yes.
10:01It is. Or then we need to make sure we know what happens if it is Jordan and also what happens if it isn't Jordan, for example. So even to yes, and this is intelligent AI.
10:11And so, basically, we can just write a prompt to say this is how you're gonna decide whether to move this person forward. And so the the prompt is the lead confirms their name matches the form or otherwise acknowledges that they applied for a service. No.
10:23If the lead says this number isn't theirs or they didn't apply for a service, but we can change this to whatever we want, and the AI will intelligently determine whether or not to to move them across. Okay? And we can also add multiple additional branches if we want to so we can make some really complex flows with the system as well.
10:40In fact, I'm gonna show you some some pretty I'll see if I can dig out a really sophisticated flow in this video so you can see how how advanced these can get. K.
10:48So now we can start going into our questions, and these are our free qualification questions. So the first question we got is homeowner. And so we wanna just adapt that question.
10:56We could just keep this open like this. It's like homeowner question mark, and we allow the AI to to effectively we move it over to creative. We allow it to just decide based on the respond to the individual or how to ask their homeowner, but I would recommend making it a little bit more specific.
11:11So are you are you currently or are you a homeowner or renting?
11:19Okay? Let's give them two options. Make it really simple.
11:23And then we're gonna go ahead and do the the budget question. Okay?
11:33Okay. I'm actually gonna add an additional question because we're asking for a budget, but we don't even know what the lead is actually asking for. We don't know what they want.
11:40And so if we just go add components, we can insert a node in between it. And so what we can do is we'll go ahead and we'll add a qualifier. So it's gonna be a text question.
11:51What work are you looking to get done? Obviously, we wanna give a little bit more thought to these questions.
12:00And so we're also gonna save settings over. We're gonna take this as homeowner, and we can do project scope.
12:12Cool. And we also want these to be required, so we don't want someone to be able to just skip past this.
12:19What's the budget looking to start? How soon are you looking for this to be done? Alright.
12:29Hit save. So these are our qualifying questions. I would recommend a maximum of five qualifying questions.
12:34Generally, the fewer qualification questions we have, the more people are gonna go through this process and successfully book a call at a higher appointment set rate. The more questions we've got, the higher quality the lead, but a lower the appointment set so in set rate. So with more friction, we get less leads with a higher quality.
12:53So now we've got another branch. Are they qualified? Yes.
12:56If the leads confirm that they are homeowner, have a project budget, and are looking to start. So is project budget above, let's say, 5 k and are looking to start, and the lead fails and one or more have been been the homeowner having the project budget or having a project budget under 5 k, and we'll put a currency there as well just so it's nice and clear.
13:21And then we hit save. Okay? We're gonna test this out in a minute as well.
13:25Now if they do qualify, then we're gonna go ahead and we're gonna book a call. And we don't have a calendar selected at the moment because we haven't integrated our CRM. We will do that.
13:32We can see the flow of exactly what's gonna happen. So we choose the calendar. We set the opening prompt.
13:37So, like, based on your answers, you look like a great fit. Let me set up a call to discuss this further. We can change that in whatever way we want.
13:43And then when they say yes, call us book, and we did a ranges of time with them based on the calendar availability. So check the calendar availability, send over some suggested times, and then book them in. My recommendation to you is try and avoid sending calendar links.
13:57If you send calendar links, you're gonna get far less conversions on the back end. So always try and suggest times. It makes it feel a lot more human.
14:05Perfect. You're set. We'll call you at this time on this date.
14:07And then after it's put in the appointment, we want it to stay active. Just in case the lead reaches out and ask questions before the call, maybe they've they wanna reschedule or cancel the call. The AI will handle absolutely all of that.
14:19So we wanna keep the AI active. We wanna have the availability and options. Generally speaking, for most service businesses, seven days is okay.
14:26If you're a high volume business, three days is where I would be setting it. I would try and avoid fourteen day or thirty thirty day bookings. You're gonna get a high show up rate issue.
14:37And we do wanna now cloud cancellations and rescheduling, and then we wanna send post booking reminders. So twenty four hours before the call, we're gonna send a reminder.
14:44One hour before the call, we're also gonna send a reminder, and we can have a look at the flow there as well if we are interested, and then we're gonna hit. Done.
14:53Cool. If they don't qualify, okay, then we're just going to we've got a objection handle. So if the leader hesitates, gen gently probe just to clarify, is it more than the timing's off or you're not sure about the budget?
15:06If timing suggest a quick call, if budget reiterate value and ask for a decision. Now I'm not a massive fan of that qualifier, so what I'm actually gonna say is if the lead is not qualified, determine the reason why bunch it budget what have we got?
15:31Tartan, which is either budget or timeline and ask them as they are will either willing to move faster or willing to increase their budget for the scope of their work.
15:56Okay?
15:59I'm gonna hit save, and we'll put that bad boy to the set test. Okay?
16:04If it's flipped to a yes, yes, the lead now wants to book a call or is qualified.
16:12No. The lead remains uninterested or they are unqualified.
16:19Okay? Even no, it's gonna mark them as lost. If yes, then jump through to book appointment.
16:26Okay? So if they are then requalified, put them into the appointment.
16:31And, basically, what this does is that'll skip through, say, to this note here. So it'll actually go back through to this step if we have the opportunity to objection handle the requalify them and book them in. So it effectively loops back around.
16:43So that is our simple agent for Meta. And let's actually just get rid of Wise one on the right hand side there. And let's go ahead and actually test this agent.
16:52Can we make this full screen so we've got more room there for you guys on it? It's coming off my my full screen. Okay.
16:57Cool. Let's go ahead and do that. So we hit test, and we can go ahead and we can test this with, like, a genuine lead, for example, on our CRM system.
17:05We're gonna hide this for now. We're gonna keep this nice and simple. We're just gonna hit start test.
17:10We'll actually see this in motion so we can follow the trace. So we got the first message sent there. Let's just get rid of that help.
17:19Yes. It's Jordan. Cool.
17:21So it's taking us through the logic. Now I'm gonna ask us the first qualifying question. Thanks for confirming, Jordan.
17:29Are you a homeowner or renting? I'm a homeowner.
17:34You can imagine these messages are being sent on SMS, WhatsApp, Instagram, Facebook, email.
17:42Depends on where you set this up. Got it. Thanks, Jordan.
17:45What work are you looking to get done? I want a new conservatory.
17:52Your conservatory is a great addition to the home. Do have a rough budget in mind? Don't really know.
17:59About 10 k. We will do another.
18:03We'll do a difficult client. We'll do a difficult lead retest shortly when we we add some knowledge into the system as well. Won't always be sunshine and rainbows of leads as we all know.
18:14That gives us a good starting point. How soon are looking to have this done ASAP? Graves, now it's thinking, is Jordan actually qualified?
18:23Yes. He is. Awesome.
18:24He's gonna go ahead and book a call, but it's probably fail at this point because we don't have that call actually, a calendar integrated with that. Interesting to see actually what this does because I don't have the calendar integrated right now.
18:37Yeah. As expected. Sorry about that.
18:38My calendar is acting up. What day and time works best for you? Presumably, the agent is just gonna go ahead, get a time, and try and mark that on the on the CRM.
18:47But, yeah, we know. Let's go ahead and actually set this up and let's actually make this more sophisticated because right now, this is an agent that just has our website and very little knowledge of the business itself.
18:58And so what we're gonna do and before we go ahead and go over to the knowledge where we're gonna give it more information, we're gonna go over to agents so we can actually see what's happening on the back end. Because those of you that wanna get a little bit deeper into prompt engineering, you'll see that the prompt has already been created for us.
19:12And so this is what we call the agent mission. There is a very simple version of this, and we have an advanced version of this. And so you can see the identity.
19:18You are James, a rep for Angling Home. You speak with inbound leads via SMS and chat. Your job is to qualify every lead for the workforce structure.
19:24Yeah. The other, this is your goal. This is the tone.
19:28This is the pricing. Okay? Do not reinvent ranges, tiers, numbers, very important.
19:32Guardrails as well. We can also say guardrail do not use Emdashes. Okay?
19:35It's something I personally don't like in messages, but AI is very hard to out prompt on that side of things baked into it. Disqualification criteria there as well, and we can have the response timing, and so we can set base in fact, I'm skipping ahead here.
19:49Language, of course, is very important. What language do we actually speak and what do we want it to speak because we can have this bilingual.
19:56What do we want the personality to be like? So we've got writing styles. We've got the sales approach, and we want it be a closer or passive voice description, example messages.
20:05So if you you're let's say you're a law firm and you really wanna make sure that this this is absolutely how you speak. You can upload a bunch of example messages, any avoid and and phrases and so on and so forth. You can get a lot deeper into this.
20:16We've just created the baseline agent, which which is just good enough to get going and book appointments, but I would recommend getting Nitzi into the background of in tweaking over time. Asking a human, if someone wants to escalate, what's the email address?
20:32Okay. And then we also have the response time. And so wait to collect multiple responses.
20:37Okay? So broadly speaking, we don't wanna respond straight away to every single lead. Sometimes someone will respond, and they'll be if you're like me, you send one message to, like, half a sentence, send another, send another.
20:47I'll send, like, 10 messages where other people would just send one paragraph. And so you gotta wait a bit of time before to allow that person to actually respond, and then you've got a wait time after collecting.
20:58And so thirty seconds, broadly speaking, is a nice wait time. It feels more human. If it's immediate responses, then it's it's not gonna be feel very human.
21:07It's gonna feel a lot awful lot more AI. And, uh, whilst we never wanted to see people, the arcs of AI appointment setting and how we get the maximum appointment set rate is through people thinking they're speaking to a human realistically.
21:21And so we wanna make it as realistic as possible on that front. I would also recommend turning on split messages.
21:27And so long messages from the agents will be broken up into multiple messages. So it, again, feels more human and real. You can turn emojis on if you want to.
21:37You can also auto extract leads data. And so if within the conversation someone has given us information on themselves, we can enrich that information and upload it on the CRM system. Would always recommend turning that on too.
21:50Emojis could be good if you're if you're messaging on WhatsApp or actually all platforms emojis can work, but it really does depend on the brand. And then we also have dynamic follow ups as well. So if somebody has not responded to one of the qualification questions, we can see how many dynamic follow ups we actually want to go ahead.
22:08Okay? And a dynamic follow-up is gonna be a completely custom follow-up based on the previous conversation. So for example and this lead had said, you know, I want a new conservatory and didn't respond to the budget, and it will follow-up with that person up to three times with things along the lines of, hey, you're still interested in that conservatory.
22:25You know? So it's a lot more custom than just following up, just bumping this like most people will go ahead and do. Okay?
22:32Now one thing I would also have a little look at as well is we can look at our we can well, we can look at, first of all, components. So we can go ahead and we can add any components into the mix here if we wanna create an up and customize this agent itself.
22:46So whether that's, you know, sending voice notes, for example, whether it's a web hook to trigger something on an automation software, whether it's transferring to a human, more different agent, whether it's adding a wait time or delay for more realism, or it's marketing leaders complete.
23:02So also and we can even add split tests. So over a large amount of data, we wanna split test whether or not certain messages, certain qualifiers are are getting us to the roots of of booking more points or qualified leads. We can also turn on follow ups here.
23:17Actually, an important distinction. These follow ups or if someone I got it wrong. If somebody hadn't responded to a a previous message.
23:25So if somebody we sent a qualification question they didn't answer, that's what these follow ups handle. The follow ups that we've got here, dynamic follow ups, these are if someone asks to follow-up. So if someone says, I'm busy right now.
23:36Message me later. Can't talk right now. Speak next week.
23:39The dynamic follow-up will handle that. These are the follow ups where someone actually just doesn't respond to our message, and those are those AI generated follow ups that I that I referred to. And we can just do a custom AI prompt.
23:51We can manually write the follow ups if we want to, but the most effective is gonna be those AI generated follow ups that I described. And so we can set five follow ups. You know, you can send more follow ups than that.
24:00I would set a minimum of three follow ups. We can also then set the timeline of that. Now you'll notice here, this is a hack.
24:06You'll notice that we've got the timeline of three hours. We recommend following up three hours. I actually personally recommend doing a three hour follow-up and then doing a six hour follow-up.
24:17So we're following up twice within the first twenty four hours and then, you know, a third time just after that, one day later, because the lead is the hottest. We . They opted into something.
24:28They got distracted. Stay on that person. Imagine, like, a friend came up to you, knocked you on the shoulder and said, hey, man.
24:33Do you wanna go for dinner tonight? And then they started ignoring you. Be like, dude, like, do wanna you go for dinner?
24:37You just asked me if wanna go for dinner. Why are now ignoring me? This is weird.
24:40Take that same mentality. That lead opted in. They're gonna be cold.
24:43They're gonna go to somebody else. They're gonna be distracted. So follow-up with them multiple times in that first twenty four hours, and, of course, cancel reply when they reply to you, then those follow ups can be automatically turned off.
24:53Great. Let's go over to the knowledge base. I'm sure you're looking at these things as well wondering what they are, so we'll come on to those in a moment.
24:59Knowledge base is really simple. We can go ahead and we can add a document. We can add a URL so we can actually scrape the website of the company.
25:06So we go on that URL, and we can see all of the information that's being scraped straight from the website. We can also go ahead and we can upload files, or we can paste text. Okay?
25:16So what I would recommend doing is just getting as much information as you've got for your business, for your client's business, and offload it into this knowledge base. And it's gonna mean that that that the agent is gonna just be more and more intelligent. Okay?
25:28So I'm gonna add another page onto this this right now, which is actually the agent's information on conservatories.
25:37And if we have another conversation in a second after it's had the chance to scrape that, then we'll be able to stress test it an awful lot more awful lot more. In fact, let's actually do that right now. Let's go ahead.
25:46Let's just refresh this, make sure it's actually taken out information. So where the URL, I can also rename this, of course. And so what is this one?
25:54I think this is the conservatory Just so we can keep that separate.
26:04Yep. And now let's go ahead and test this. Let's test this agent.
26:15Let's say I'm a homeowner homeowner. We will be kind for now, and we'll start asking some questions now.
26:22Okay. Don't know what kind of conservatories do you do?
26:28K. Look. Stane.
26:29The leader's asking a question about a business. I just flashed up and disappeared very quick.
26:35Just gonna look into the knowledge. We enter a wide range of styles, some Victorian designs to model lean conservatories, how we narrow down more specific home improvements or additions. Kinda leaning towards Victorian, but don't want it to be old as shit.
26:51Are they expensive? That makes sense. We can definitely do a Victorian style with modern materials so it looks fresh rather than dated.
26:59Nice. Nice little objection handle there. Pricing really depends on the size and specific features you want.
27:05So to give you a better idea, what's the budget? Honestly, I'm broke. So, like, 4 k.
27:12Thanks for being upfront about your budget. Since our projects typically start 10 k, we won't be the right fit for you right now, but I wish you the best of luck with your own improvements. So that's really interesting.
27:22So now we're being de qualified immediately because in the back end of the prompt, we've got dequalification is gonna be 10 k. Now we what we do have is we've got this right down here.
27:33It's like, is the lead actually qualified? Well, then we're gonna try and requalify them. But what we actually need to do is change the back end prompt to say, don't de qualify someone until they hit the qualified branch.
27:47And then basically, what we need to do at this stage and so what I'm gonna go ahead and do is I'm gonna go into the agent. And and this is what you would do over the course of your first, like, day or two of running your agent. It's really important that you go ahead and and let's just get rid of that for now.
28:07It's really important that we we go ahead and we test this out so we make sure it's exactly how we want it to be.
28:15Okay? And we go ahead. Let's go on conversation builder.
28:20Let's go ahead and test this. I'm gonna bring this back to the the that stage in the conversation so you don't have to watch all that again. Okay.
28:28So now we're at the budget question. That's just before k. Hopefully, we will not be dequalified, and we will just pass straight through unless we've missed something somewhere.
28:40Come on. Decision pass. Awesome.
28:44So let's continue being an annoying bastard.
28:50How soon you looking to get done? We will say ASAP, but let's see if because we're not gonna be qualified, it should take us through to this branch, and then it should ask us to increase our budget.
29:03Nice. There you go. So let's see what we get asked.
29:05Since I've brought project Sibilysar on 10 k, you're open to increasing the budget to match the scope, would you prefer to adjust the timeline? Interesting. So what what we definitely need to do is change the prompt here and ask them if they and ask them whether it's budget under and ask them if they are willing to move faster or willing to increase their budget.
29:24So it kind of is worded in a way that says, ask them both of these things. What we need to say is ask them one or the other, not both. But for us, we'll say, no.
29:33That's way too expensive. I don't want it anymore. Actually, we'll be we'll be kind because I want it to then loop back for you to book a call.
29:48No. But if it's shit, I I will knock it down.
29:54The lead has agreed to increase the budget, so we should go, yes, back to book a call. Awesome. So little bit of prompt engineering, a little bit of tweaking, a little bit of testing is gonna go a long way to get the agent exactly where you want it to get to.
30:08Now I'm gonna try and dig out a really complex agent because this is quite simple right now. And, actually, if you guys want me to share this agent with you so you can save the time that I just spent, albeit not so much, in building this agent and you want an agent you plug into your age your clients, but also to your own business, I can go ahead and I can copy this workflow, and I can share it with you guys along with a setup guide for a point wise.
30:31And all we need to do to get hold of that is comment underneath this video. Let me know where you're at right now and what you're looking to use our agents our sales agents for. Subscribe to my channel.
30:40Make sure your notification bell is turned on, and DM me on Instagram, Jordan Platton. There'll be a link in the description.
30:48Send me the keyword workflow, and I'll send you this exact workflow and the setup guide. Although I won't send it to you, Pointwise is gonna send it to you for me.
30:57Right. Let me find a complex agent for you. Okay.
30:59This is gonna be a big one. So I was just sent this by the client success team that I built for a home improvement business that I mean, look at this.
31:07This is a very sophisticated flow. Now what this business does is it actually segmented this flow and generating leads on from go high level, presumably from.
31:19They're confirming the person and then they're saying, which project are you most interested in? And they've segmented each individual service into an each individual flow.
31:27So so kitchens, bathrooms, flooring, joinery. If you're a dentist, this might be implants. It might be Invisalign and so on and so forth.
31:34And then there's an individual appointment setting flow for each of the individual services. So it's hyper personalized for every single lead because every qualification question would be different depending on whether or not they need a kitchen fitting or they need a conservatory or whatever that may be. And when you're dealing with hundreds upon hundreds of leads on a monthly basis, this is how you get those small incremental gains on conversion rate on the back end by giving people that really personal experience.
32:00Now this is not the kind of agent you need to go ahead and build out when you're first starting out. And, actually, there is a done for you team at a point wise that that build these out for people. It's the done for you team that sent me this over as an example for this video today.
32:13I'm interested, actually. I wonder if we can see some analytics, if we can see some of the the the data on it.
32:19Oh, we can. I have to make sure we blur out any editor.
32:23Please make sure we blur out company and and and so on. And so there's no yeah.
32:28And there's no and there's no information, but, hopefully, I can show this. Well, 560 total leads, 183 active that people that are currently in conversation, and 36% win rate.
32:40So 36% of leads in this flow. I've actually gone ahead and booked an appointment, which is pretty amazing.
32:46Nine hour average conversation. That's very interesting. I'm not sure what's happening there.
32:50But and a 175 loss. But regardless, 36% win rate from the leads that have actually gone ahead and respond to this flow.
32:58It's absolutely crazy. And you can go ahead and look at, like, the individual flow within that and the journey that people have actually gone ahead and taken. So that's as complex as it can get.
33:07I've also got a pretty sophisticated flow set up on my Instagram for all the various keywords that people message me for to send over various resources. That's a pretty hefty one as well, but I think this tops it for for for for the top list of the ones that I personally see on the platform.
33:21Right? Let's go back over to the other flow that we would we just created the simple one, and we're gonna go through the integration steps. We're actually setting this up on a CRM logo high level.
33:30Cool. So what we need to do to get set up is hit integrations, and we can go ahead and we can just add our Google Calendar.
33:36First of all, if we don't wanna set up on a CRM, we can add Calendly. As I said, we can do Meta. We can do Twilio, and it's very easy.
33:42We just hit connect and just go ahead and ask us to actually log into our Facebook profile, we would hit continue and give it access to that. And so, you know, you're integrated in a couple of seconds. But instead, we're actually gonna do GoHighLevel, which is a bit more of a sophisticated setup.
33:55GoHighLevel was a full CRM. If you don't use GoHighLevel as a CRM, you can go ahead and you can open Zapier there and connect your CRM that way. So we're gonna hit connect.
34:04It's gonna ask us to log in to our account. And whilst we're actually doing that, for those of you that aren't familiar with Go HighLevel, let me just quickly load up their core landing page. I'll also put a link in the description to a very detailed video where I go through exactly how you can set up the Go HighLevel account and make the most of it.
34:20Go HighLevel, well, now identifies as an AI powered business operating system. I think that's actually a pretty fair repositioning high level if if any of you guys ever see this. But a high level is an all in one client relationship management system that basically replaces a shit ton of other tools.
34:35Okay? So you have spots in your active campaigns, your website builders like Wix, your ecommerce, your email marketing, your your appointment booking, your your AI voice agents even there as well, ad management.
34:48So it's just got so many different things. It's basically like a gigantic roll up of so many different softwares consolidated into one for one monthly fee, which is why I love it and why so many other agencies absolutely love this tool because it just saves us an awful lot of money. It saves us headache coming to log in to all these different softwares all over the place.
35:07So let me just log in to my account. So now we're logged in to HighLevel, but just to confuse you, you won't seem to go HighLevel logo in the top left, at least on mine, because I've white labeled the software, meaning I can put my own logo on there. So if our clients log in, it feels like it's our own piece of software.
35:21Now once we are logged in, it's gonna give us a pop up and say, I want you to give me permission. We hit next, and let's just use a demo account.
35:31There you go. Let's just use this one here. We're gonna log in.
35:35It's gonna activate us. Cool. It's gonna close.
35:40Great. Now the next thing that we need to do is go ahead and import a snapshot, which is basically Go HighLevel's way of copying settings from one account to another. And so we import that now, and then we go back to snapshots.
35:53We're gonna go over to subaccounts. We're gonna search for the accounts. We are then going to hit these three dots, and we're gonna put manage clients, and we're gonna have actions, load snapshot.
36:09We're gonna select that snapshot that we just imported, which I believe was a Pointwise official guide level snapshot 2026.
36:21We're then gonna go ahead, and we're gonna select all. We want everything to import. So this is actually just importing settings.
36:26We select all, and then we proceed. And we go confirm.
36:34And it's now gonna add all of these settings to this account. We're gonna close that. We're gonna hit switch to subaccount.
36:41So if you imagine subaccounts are kind of like all your clients, your agency account is your agency itself. Now we come through to the GoHighLevel dashboard. Now I'm not gonna get stuck in all of the intricacies of how to use HighLevel.
36:52As I set up a link in the description, you can learn about how to use it. We'd create another hour long video going forward with details here, but I highly recommend you watching that. We're gonna go over to settings, and we are gonna go down to our custom fields, I believe.
37:09No custom values. I was right the first time. There we go.
37:12Now what we need to do is add our Pointwise API key and our Pointwise client key, and we enter them into these fields. So we hit edit them, and we go back over to our Pointwise, and we're gonna go ahead and take it there.
37:24And we're gonna paste it here. Gonna update that. And we're gonna go ahead and take our client ID here, and we're gonna paste it there.
37:36Great. Now all we need to do is go back to Pointwise, and let's refresh that page, and we are all connected.
37:44So just so you know how this all works in the Pointwise side, if we were to come out of settings right now oh, what have I done there? Made that menu bar really small. Okay.
37:54There we go. We go over to automation. We'll see that the point was actually imported a workflow, and workflows are really I go high levels automations.
38:04And then we have a look at that. We go on trigger AI. These are how you know, enable auto save.
38:10These are how we are essentially effectively triggering our AI agent inside of Appointwise. Okay?
38:18And so when this happens, basically, turn on the AI. That's basically it.
38:23And so for us, if we go back over to our agent, and we made this one here, and we want this now to be a source go high level, and the event is has changed that event as as as a tag, and we have it as new meta lead. Okay?
38:42We can set it up on GoHighLevel that when a new lead comes in from Meta, they get a tag.
38:49Okay? And we want the trigger here to be contact tag, Meta lead.
38:57And so it's gonna be when a tag is added to a contact, and that tag is meta lead.
39:05Okay? Save trigger. And it's the same as what I just did.
39:10No. I did new meta lead. We'll just do meta lead then.
39:13There you go.
39:16Get rid of that one. So now we've saved that. We've got the trigger set up here.
39:21Then in go high level, we would wanna just make sure that we would set up a new trigger which actually adds this tag to contacts. So how would we do that? We'd go create workflow.
39:30We'll start from scratch. The trigger would simply be when and we'd need to create our connect our Facebook account, but it would be either depending on whether or not we've used to go high level form when a form is submitted or a Facebook lead form is submitted, okay, then we can add a filter, the form must be x form.
39:50We must build one now. Okay? When that form is submitted, then we wanna add a contact tag.
39:55So that's all we need to have. When form is submitted, add contact tag, we hit publish, save, and now we have the full flow between go high level and point wise.
40:06Okay? And then you want to know how you can integrate Meta on the back end, you just hit settings, and you go down to integrations, and you would log in to your Meta account here, and you would connect in there.
40:17But as I said, I go for this in much detail in a in a future video. And so, yeah, that's exactly how workflows work. You don't even need to to worry about workflow too.
40:26It is all done inside that trigger AI. Now one thing we will need to do is create a calendar because we wanna be able to integrate our calendars.
40:34We go to calendar settings, and we go ahead and we create a calendar. We can say personal booking, test, and we're gonna hit confirm.
40:45Let's select a team member. We don't have a team member. Let's create a team member.
40:50Go to my staff, add user. We save that person.
40:56Now we've gotten them. We go back. We go to calendars, go to calendar settings, create a calendar, do personal booking, hit tests, setting a team member.
41:05There we go. Hit save. Now we've got a calendar made.
41:08We can go into a point wise. We can select what calendar we now want to actually integrate. We scroll down appointment booking, winch calendar, test.
41:16There it is. Hit done. Hit publish.
41:19Hit check for conflicts. So the AI is actually gonna go ahead and search. If there are any errors across this agent, no conflicts found.
41:29Push live. And now this agent is completely live. One dead end of appointment.
41:34Okay. What is it? Oh, that's disappeared.
41:38I can look into that. And then we save. So now we have a full end to end meta form submission qualification appointment booking agent built out in what are we on?
41:50Are we under an hour? Just over an hour? Maybe?
41:53Okay. Cool. What else do we wanna look at?
41:56What else do we wanna see? And there's some really cool things that have been built on Appointmentwise over the past year. One of those things being voice is pretty awesome.
42:04So Appointmentwise now integrates with Eleven Labs, and so you can go ahead and enter your Eleven Labs API key. So if we hit add API key, I should be able to take one of mine. Okay.
42:14Let's just go ahead and take one of those. Let's just copy the key, paste it into a Pointwise, connect.
42:23Okay. Now we've got a number of different voices, and we can select from those voices.
42:28We can even do a custom voice there, and we can test and play around with the settings. We can do a little test. We can preview it.
42:33And so we can send voice notes to people inside those those agents, and so we can send automatic follow ups via voice. K? And we can create snapshots, which is what you're gonna get access to if you comment on this video and DM me.
42:45I'll give you a snapshot of the agent we created today. We have full analytics. We've got all of our contacts, which we won't have because this is a new account.
42:52We have all of our conversations that are stored in here as well. So regardless of whether it's from your CRM, all the conversations will be stored in here. You can even take over conversations as well, and humans can take over.
43:01One thing that I didn't like in the first version of AppointWise is if your client started messaging a lead, then they find themselves in this battle where the AI was trying to respond to the lead, and the client was trying to respond to the lead, and it was it was one big mess. And now if a client responds to a lead, then the AI stops responding, and and so they can do so in here if they want to.
43:19Full pipeline management. So if and there isn't a CRM system, you can go ahead and you can create a pipeline and manage all of your leads across that pipeline as well.
43:28And my personal favorite, the self learning engine. Because now there is WiseOne, which is the large language model that AppointWise runs on. With every single conversion on the platform, every single lead, the system gets more intelligent, and your agent and your individual account gets more intelligent.
43:41And so all of the knowledge that is stored, any of the agent learnings, Ellie, any unused knowledge, any failure patterns, any success patterns, it's gonna learn over time, and it's also stored in this really nice graph database, which is pretty average there. There's not really any information on there at the moment because we've just gone ahead and and set up that, but we can also see how that information relates across the board, and we can drill it down into different insights.
44:03But maybe I can go into that in in more detail in a future video. It's really cool to see as you speak to thousands of leads, how all of that data is connected and how the system is actually learning.
44:15And, of course, if anything goes wrong, then you can go on to diagnostics and it'll audit your account, and you can see if there are any issues there as well. Hey. Get help.
44:22Speak to WiseOne for any support, or you'll see that it'll help button in the bottom right, which I actually got rid of earlier where you'd be able to get straight through to their support too. So that is how you build text based AI sales agents, the full a to z. It's been a pretty chunky video today, but you no longer have to hire SDRs or appointment centers and pay them $3,000 a month only for them to complain, for them to get unmotivated, and then for them to sleep.
44:47How dare they sleep? Now you can pay a robot who doesn't sleep under a $100 a month who can handle thousands upon thousands of more leads than they can twenty four seven in any language across any industry, and they'll get a better conversion rate because they reach out to leads immediately.
45:01Did a study on his own leads. He found that they increased their lead conversion rate by 21 times when they reach out to all those leads within ten minutes. Humans can't keep up with that.
45:11What a time to be alive. Subscribe for more, and I will see you very soon. Cheers, guys.
The Hook
The bait, then the rug-pull.
The pitch is blunt: agencies and business owners have been paying $3,000-a-month sales reps to chase leads that came in through ads. This is a live, full screen-recording build of the AI agent meant to replace that job for under $100 a month.
Named ideas worth stealing.
03:10model
The Branching Qualification Flow
- Intro + identity-confirmation branch
- Qualifying questions (max 5 recommended)
- Qualified/not-qualified branch
- Book appointment, or loop into objection-handling
The core structure every agent in the video is built around: confirm who you're talking to, ask a short set of qualifying questions, branch on whether they pass, then either book the call or route into an objection-handling loop that tries to requalify them.
Steal forany inbound lead-intake flow, sales DM funnel, or booking form that currently has no qualification step
27:30list
Two-Touch 24-Hour Follow-Up Cadence
- ~3 hours after no response
- ~6 hours after no response
- one more touch the next day
A specific follow-up schedule for leads who stop responding mid-qualification, timed to catch them while the lead is still 'hot' rather than letting a full day pass with silence.
Steal forany lead-nurture sequence, cart-abandonment flow, or no-show recovery sequence
13:40concept
Suggest-Times-Not-Links Booking Rule
Instead of sending a generic calendar-link, the agent proposes specific times based on calendar availability and asks the lead to confirm one, which is presented as converting at a meaningfully higher rate.
Steal forany booking flow, whether AI-driven or manually sent by a human rep
How they asked for the click.
VERBAL ASK
30:45product
“Comment underneath this video, let me know where you're at right now and what you're looking to use our agents for. Subscribe to my channel, make sure your notification bell is turned on, and DM me on Instagram, Jordan Platton... send me the keyword workflow, and I'll send you this exact workflow and the setup guide.”
A layered ask (comment + subscribe + DM a keyword) that also gates the free workflow-and-setup-guide giveaway behind engagement, executed mid-video rather than saved for the outro.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
AFFILIATECommission earned if you click.
OTHER LINKSAlso linked in the description.
Visual structure at a glance.
