Introduction

This is me, so I'm a fractional CTO at Gencut, video AI editing company, and I build AI agents.

Social Proof and Personal Credibility

So, what makes me great, Google Cloud asked me to work for them. On average, I have between 117 and 200 people in my events on LinkedIn without paying any euros for marketing. And I went viral on my account and multiple accounts of my clients.

The Michael Scott Lesson: Sales Is How You Make People Feel

So, does anyone know who this guy is? Yeah, who you see? Michael Scott.

And why is Michael Scott so great at sales? He's a people person. Yes. And what would you then say is his strongest skill?

Being awkward. Empathy.

The Chili’s Meeting: Relationship Over Pitch

I don't know if you guys remember this episode, Chillings Meetings. Basically, that's an episode where one of the most important customers... they're not sure if he's going to renew.

I think you don't see that well now with the light.

So his boss, Jen, she is like, Michael, Michael, please close. Please talk about sales. And Michael says, no, we're gonna go to Chili's, we're gonna go to a restaurant, and we're gonna give him the best time of his life.

He's gonna eat great, and at the end of the meeting, he goes to the guy, and he just sells them immediately.

Why Feelings Beat Features

And this is the most important thing about sales, it's all about how you make people feel, not about how much value you give to people. So I wish it was actually about value, because then I would have closed many more people in my career, but it's all about how you make people feel.

Trust, Bias, and How We Form Perceptions

Why is that? Why can't you just not sell based on value?

So it's because of this.

Early Trust and Confirmation Bias

So this girl, she 100% trusts her dad and her mom. She doesn't know. And the reason why she doesn't know is because if you really take the time to know someone, they trust you.

Could be a coke addict, could be a murderer. It's basically confirmation bias. And this is why it's literally programmed in your genes to listen to someone when you really are with them.

From Emotions to Personality

Then, when you grow up, you learn different things. You learn, actually, emotions. Stress, anger, anxiousness.

And we learn that certain traits which our parents actually loved in us were not the traits that other people love in us. And from this, we create personality. this framework.

Archetypes: How Others See You

So if you were wondering who I am, I'm this guy, the debater, same as Steve Jobs.

And this is not who you are, this is how others see you. And up until you don't know how others see you, you will always not know who you are.

Because people will always assign you in your mind a certain archetype. because that's how they relate to you.

Live Demo: Personality Profiling from LinkedIn

To showcase that this is just not like fantasy and just not one of the models, I'm gonna actually test this with someone in the room here. Wu is willing to share their name so I can search him on LinkedIn and then I can tell stuff about it.

Who wants? Yeah? So what's your name?

Wait, I will just let you type if you come here. It's just faster. Perfect, cool.

Method: Profile, Posts, and Comments

So first thing I want you to do is you first save the PDF because your profile is what you would like people to think about you. Just the PDF is not enough. You also need what you say about yourself.

So what your post is, what you think is most important to share with the world. So what you have to do here.

Yeah, but that tells me like, I see a lot of events, interesting. Really wonder what kind of person will come out of the thing.

So you take all the posts, put it in here. Doesn't matter which language it is.

I'll go to comments. But comments is where you actually find the truth. So in comments you can never hide.

I'm the debater, so I always ask the hard question. You know, that direct thing. And in the comments you can also be like supportive, or you can be very lengthy.

And if you combine these things together, let me show you. Nice.

AI-Assisted Typing and Sales Angle

Just for people who notice, this is more than just a custom GPT. It also has a book about personality, so it's better in identifying who you are. So let's see what it says.

So according to this, you're an INFJ, which is the same as my sister. So now let me actually check this.

So you're extremely social. It took you a very long time to find a boyfriend if you have one at all. Is that true? Because you're very picky, if you're honest.

You're laughing because it's true. You are the organizer in your friend group, but you're not the organizer as in that you're always super direct, but you are kind of like the leader. the social leader.

You care a lot about how you look like. Every detail is like picked in your life.

Okay, okay.

Translating Personality Into Sales Tactics

Again, it's not 100% perfect, but most important here is about the sales staff. So is this true? If someone were to sell to you, you would do this?

So use emotional resonant language, acknowledge your role in the community building, provide vision, purpose, and future impact. Okay, let me go further. Are these the questions that you would ask in the proof and credibility signals?

Preparing for Sales Calls with AI

Yes, so as you can see, so this is a simple example. I, before I go in any meeting, I always take the personality, and then I know like, okay, this is the role where I have to step into.

Simulation and Objection Handling

Even if it's more important, I will simulate you, so when I have a hard conversation, I've had it 20, 30 times. Every question and objection that I can have, I've heard before, because I know, you know, I can reprogram you as a clone in JGPT. I'm not going to show that today, but I just want to show you how far you can go with AI.

Modern Sales Philosophies

Sorry, to get back to sales, because we are here today to talk about sales.

The Hormozi Speed Doctrine

So everybody knows this guy. So this is Alex Ramosi. He's the epithet, the ultimate man in sales.

Basically, his philosophy comes down to like five things. And to summarize it... It's actions per minute, productivity, and if you go, if it works, you just increase it.

Be very fast. And also if it doesn't work, you just try a different way.

Know who you're talking to. You keep clicking on that button, yes?

Check on LinkedIn.

From Volume to Precision: Super-Targeted Audiences

So I believe that it's actually going to be different nowadays. A great mentor of mine and friend of mine, he said, it's not about anymore how many people you have, it's actually about having a super targeted audience. And what do I mean with that?

Proof, Questions, and Follow-Up

So what I showed you to you, you need to have a kind of proof, questions and follow up, so you understand the customer, so you know which role you have to play in order to connect to them. As I was showing, you first have to make them trust you, and then you also need to have enough data to really say that's my customer.

how many times I talked to CEOs and founders and I said like, why is that your customer? Can you show me the data on how much traffic actually from that segment is going to your website and is converting?

And I was just like, we always do it like that. Our customers are like that. I'm like, prove it to me.

Building a Data-Driven ICP

Make data-driven ICP and recognize then where your customer's coming, where they're going to, and how do you do that?

Why Clay: 30+ Data Sources for Deep Insight

We have Clay. So Clay is a tool. What is so great about Clay?

Clay aggregates information of not one, not two, but 30 databases. We have so many data points that you can truly, deeply understand your customer. And what do I mean with this?

Signals: Hiring, Tech Stack, and Intent

So this is a company, a finance company that I work for, and we were uploading all their four to five clients. To do this well, you need their website. And then you can look at what kind of jobs do these companies look for? Because if you can see what kind of jobs they're looking for, you know, like, oh, they're actually looking for something in finance, or they're looking for something in marketing.

We also now look at the tech stack. So we see on average that most companies have between 70 and 30 sort of kind of like platforms they use.

But if they use, for example, WordPress, you're like, oh, they don't care that much about innovation, or at least they use like outdated CMS, according to me. I think Webflow and Framer is just like much better, or you code your website yourself.

Yeah, I agree, thank you.

Synthesize Insights and Define Anti-ICP

And then, I call it, you aggregate all these data points, and this is like, this is the first row, but after this, there are like 30 to 40 data points. You put them in ChatGPT, and again, and ask like, what can you tell about my customer, what I didn't know?

And you'll find very surprising insights, which afterwards you can create a target audience, but also the anti-target audience, anti-target audience. So it's the people you should not talk to.

Simulated Users: Faster, Cheaper Customer Research

So, and then you take it one step further. So this is what I do, I also create AI strategies for companies.

Anyone here has ever scheduled customer interviews? Yeah, takes a fucking shit ton of time to get people to actually talk about it.

So instead of that, there's something which we call simulated users.

Create Segment Prompts and Converse

So simulated users is when after you have created this data-driven ICP, you generate of every customer or segment a prompt which simulates them in real time.

I've done this for multiple people, and afterwards you can then ask them in real time 20 questions. Make sure to ask the one question at a time, like a real conversation, so it's not like 20 questions and then you get like an output back. You really have to create a conversation.

Cross-Model Validation

Do this with different large Magnus models, so Claude, JTPT, Gemini, who have different results.

And yeah, I always learn a lot of new things which I didn't know before.

Sales Agents in Practice

Okay, so because we were talking here about LinkedIn or about sales agents, one of the sales agents that I built is like this profiler.

So basically, actually I think, do you guys hear me good enough without this? Okay, I'll do it without it, that's easy, cool.

LinkedIn Profiler Agent

So this is just a LinkedIn agent. It goes to your LinkedIn profile and it checks your post and your profile and then based on that it tries to analyze what your MBTI is.

I think this one is also easy.

1% Better Every Day: Coaching and CRM Automation

So if you wanna be good at sales, you always have to get every day 1% better.

And you do this by taking the meeting friends crowd, then afterwards you summarize that, then you make sure it goes through your, how you call it, CRM, because no salesperson ever like to make notes. And yeah, it's so true.

And then you make sure that the salesperson gets an AI sales coach, which every day is like, oh, did you use the right sales framework? Did you listen to the right objections? And did you ask the right questions?

From Transcript to Proposal

Okay, then to get back to yours, yes, we of course use NCP and stuff.

After you have a transcript, you throw it to an agent, the agent then creates this proposal.

I'm still waiting for the Figma and Canva API, because those are already available, but only on enterprise level. And for some reason, I don't get why they don't make it under that. But yeah, that would make a lot of sense.

Next thing.

Live Build: Workflowing an AI Agent in Relay

So instead of talking about this, let me just show you how easy it is to actually build an agent. So if you go to the Relay app,

Okay, so here we have the Relay app. I will make a new workflow.

I can also see how fast my internet works.

Inputs: Transcripts via Fireflies

So let's say I would like from Fireflies, which is my... So for everyone, there are two big transcribed tools, Phantom versus Fireflies. I'm a big Fireflies boy, and the reason why is because they have a very good API documentation. Phantom has no free available API, but you can send a webhook to make.

So first you click on Fireflies, you click on Friendship Generated, and you say like, oh, it's not connected. You just go to Fireflies.

Then I go to the next step, which is like, I want... I want an AI, summarize, okay.

Prompting Strategy and Content Generation

So then it has access to the transcripts and I will, if you're ever gonna create an AI agent, I never ever write the prompts myself. You always say like, you are the ultimate prompt engineer. Doesn't matter if you actually are good at typing, it will just fix it yourself. And then you say like, I need the prompt.

And then you specify normally the tools, but because we are only doing this and now you don't have to specify tools, I need a pond where it uses the transcripts. And then I want you to create a LinkedIn post. Give me ideas for what kind of format.

And then rank on viability, on viral, viability, if it can go viral, if it can go viral, why this would work with my audience. Let's say people who want to learn AI agents.

So now it generates this thing. As you can see, I'm horrible at spelling, horrible at grammar, but luckily there's something like ChachiPetit, Grammarly, and thanks for God that I actually created this, because now I can actually operate in the world.

So we create here this. Actually, the better version to do this, it's called content stealing. Basically, you go to LinkedIn, take any content creator, whatever was good at what you do, and you just pick all their posts.

You turn them again to ChatGPT, say, make this into a template, and then you use that template to then generate content. It's called, okay, so we do this.

Go back here. Okay. Is it already connected? Can use this model. Done.

Output to Google Docs and Operational Notes

And then we go to Google Docs. And this is like why I love Google so much. It's by Google, it's just one click and you're ready.

Just making now, not directly to, oh, sorry. Oh, sorry.

Yeah, so different as Facebook. So for everyone here who's ever gonna code, Meta's documentation, horrible, my God. And the reason why Meta's documentation is that horrible is because they got caught stealing documents. So now they have like 10, 20 menus, like, are you okay?

And Google was never caught stealing stuff. You can say here, I want the AI summary.

Now you have to, of course, make sure that the stuff is in there. So we add plus data.

And then you can, if you want more stuff, I will just do it like this. Done. So I'll put some stuff here.

Title, content, oh sorry. Content ID. Normally I would allow the AI to generate multiple data points, so then you can fill them like content title, but I just, because I want to be fast, okay.

And I have to of course pick a folder. Done. Cool, and now I can start a test run.

And why would you like to do this?

Turn Every Conversation Into Content

So, if everyone here wants to sell, every conversation you ever will have with a lead, their problems, their objections, their questions, are perfect content. Because if one person has it, 100% someone else has it as well. So, always let customers sell to other customers.

So let's take this. It's not a really boring thing, but just for an example.

Stack Notes: Relay vs. Relevance AI

I built myself, my agents usually on relevance AI. This is relay, but I just want to show you this is the easiest one if you want to build something.

So in like 30 seconds, six seconds is finished. Let's go to documents. And ta-da, we got here.

a content ID. It's not that great, as in you could have more text and stuff, but yeah.

I'm not always at the best with my CFO.

Cool. Let's continue.

Interactive Quiz

Okay, so everyone here, open your phone. Now you actually have to take your phone.

It's six questions. Oh, sorry.

Is it possible to dim the light so the screen is more active? It's readable. Okay, it's readable. Okay, cool.

Please join. Take a good name. Sorry.

Are we all in? Cool.

Okay, so there will be six questions. You can also read the code on top if you just wanted it 6-6-4-6. The faster you respond, the more points you get.

Here are some additional rules. Sometimes there's multiple choice, but most of the time you just have to pick an option.

Okay, 20 players, let's go. You see the question on your screen. Sorry, the faster you react, the better. Okay.

Nice. Let's go to the next one. Let's go.

Is there a prize for second or third place? No, no, no. It's only winners.

But as I say, go as fast as possible. But as you can see, I call it, in between you will get the scoreboard so you can see how far everyone is. I think it's after this question. So you can see if you, nice.

Oh, so here you can type multiple keywords. So these were the keywords you could have typed.

Okay, so at the moment, Shammy is winning. There are still questions coming. So Watson and Wolfmeister are not that far away.

Let's go. Last two questions. Last question. Oh, sorry.

Final Question

What's the biggest leverage of AI?

Sorry, I made this really fast. Okay, so Wolfmeister, oh wait, yeah.

Conclusion and Call to Action

So, in summary, my name is Jasper Huys, I make AI agents, I am a fractional CTO, I work with these kind of customers, and if you want to connect with me, next week I'm doing a sales round table on LinkedIn, so feel free to welcome, we've got some real good speakers.