Why only 2% of AI users save 100s of hours (and you are not)?

You use ChatGPT for emails and summaries, but the results are average, and it still hallucinates. So how are others saving thousands of hours and running entire operations with AI agents?

The secret isn't in using AI—it's in building with it.

Right now, 98% of professionals are stuck as "AI users." In this episode, Luis Salazar reveals the simple path to join the 2% who are "AI makers." This leap is no longer a "nice-to-have" skill; it's rapidly becoming the new standard for hiring and professional advancement.

Learn how to shift your mindset from AI passenger to AI driver and build a custom AI teammate that understands your work, remembers your context, and amplifies your expertise.

In this episode, you will learn:

The 3 simple components of an AI agent (Persona, Knowledge, Tool).

How to build your first agent in less than 15 minutes using tools you already have.

Why "AI agent builder" is already a critical hiring filter.

The "mini-agent" strategy: Why you should build a team of specialists, not one "super-agent."

How to treat your agent like an apprentice to improve its performance daily.

🎙️ All our past episodes  📊 All published insights | This podcast features AI-generated voices. All content is proprietary to AI4SP, based on over 1-billion data points from 70 countries.

AI4SP: Create, use, and support AI that works for all.

© 2023-26 AI4SP and LLY Group - All rights reserved

Transcript

ELIZABETH 0:15

Okay, Luis, people keep asking, how on earth can you run a company with 50 AI agents when ChatGPT can barely summarize a document without making up things? What's the actual secret here?

LUIS 0:27

Oh, do you mean what is the secret sauce to manage a global operation with a workforce of AI agents? Okay, let's talk about it.

Why Users Stay Stuck At Summaries

ELIZABETH 0:35

Welcome everyone. I'm Elizabeth, your virtual co-host, and as always, our founder, Luis Salazar, is here. And today we're pulling back the curtain on that very question.

LUIS 0:45

You know, after we talked about Agent Ada last episode and shared how policymakers in four regions saved 3,000 hours in six weeks, everyone's saying the same thing. Luis, I use Chat GPT. It hallucinates, it forgets context. How are you getting those results?

ELIZABETH 1:04

Oh yes. And now everyone wants their own agent Ada or their own Elizabeth.

LUIS 1:09

Yes, everyone wants an agent for sure. For example, last week I was at Chapman University in California, room full of business leaders, faculty, and students. The number one question: I use Chat GPT every day, but how do I build my own agent? And if I build one, does that mean I'm cheating at school or work?

ELIZABETH 1:29

That question isn't really about technology, though, is it? It's about permission, about understanding what's even possible.

LUIS 1:36

Exactly. It's the question 90% of people using AI are asking right now, but don't know how to answer. It is also why I focus so much on sharing what is possible, to inspire them, to unleash their creativity for good.

ELIZABETH 1:52

So today we're answering it. From user to maker, how to build your first AI teammate, why it matters, and what changes when you make that leap.

LUIS 2:01

Here's the deal. Most people use AI for summarizing articles, drafting essays, emails, or documents, but they're stuck at that level.

ELIZABETH 2:10

And since over 90% use the free versions and are misinformed by the terrible marketing done by leading providers, they think AI is just for emails, social posts, and summaries. No wonder their results are disappointing. So what is that next step?

LUIS 2:24

Their next steps is creating their own agents, building their own Elizabeth or Ada, going from someone who uses AI to someone who builds and manages AI teammates.

ELIZABETH 2:35

But Luis, our data shows only 2% of AI users are creating agents. That's a massive gap.

Hiring Signal: Show Your Agent Portfolio

LUIS 2:42

Exactly. And let me tell you why that will change quickly. In California, I had dinner with Yasser, the chief operating officer at a leading global digital marketing firm. He told me that on an almost weekly basis, advances from OpenAI make him rethink his entire business. Weekly? That's relentless. Weekly. He and his team keep creating more agents, automating more workflows. And here's the kicker. He said the secret was grassroots adoption. It happens bottom up and everyone is expected to participate.

ELIZABETH 3:16

And this is affecting how they hire now, right?

LUIS 3:19

Absolutely. Like a growing number of firms, they won't even consider candidates who don't show initiative with ChatGPT or similar tools. They want examples of what candidates have used it for. But ideally, they've created agents.

ELIZABETH 3:34

So this is already a hiring filter, not something coming in the future. And here we are, still banning AI at school instead of guiding its adoption.

LUIS 3:43

It already is. Four out of every 10 knowledge workers' job postings list AI skills as a requirement. Students who show up with an agent portfolio aren't just competitive, they're differentiated.

What An Agent Really Is

ELIZABETH 3:55

You also mentioned a neuroscientist who approached you after one of your California talks. What did she want to know?

LUIS 4:01

Oh, that was Dr. Pak, brilliant mind and entrepreneur. We were discussing AI applications for neurodegenerative diseases. And she said, I need to create my chief medical officer agent to brainstorm with me. What's the first step?

ELIZABETH 4:17

So it's not just students entering the workforce or business leaders scaling operations. It's domain experts wanting to amplify decades of expertise through AI.

LUIS 4:28

Exactly. Three different people, three different contexts, same question. How do I go from user to maker?

ELIZABETH 4:36

All right, let's answer that question. But first we need to clarify something fundamental. What actually is an AI agent and how is it different from just opening ChatGPT and typing a question?

LUIS 4:48

Great question. Most people are having one-off conversations with Chat GPT. Every session starts from zero. But you see, an agent is different. It's like having a teammate who remembers your context, understands your work, and has access to relevant knowledge.

ELIZABETH 5:06

So it's not a tool you use occasionally, it's a colleague you've actually onboarded.

LUIS 5:11

That's exactly right. And the good news, building one is way simpler than people think.

The Three Components To Build One

ELIZABETH 5:16

How simple are we talking? Because what enterprises report sounds expensive and lengthy?

LUIS 5:21

Not at all. It's just three components. And it takes hours, not weeks. Let me walk you through it. First, you need a persona or system instruction. This is like writing a job description. You define who is this agent, what's its role, how should it communicate.

ELIZABETH 5:38

So if I'm building a research assistant, I define its expertise, tone, boundaries. Basically, what makes it uniquely suited to my work?

LUIS 5:46

Exactly. Second, you give it access to a knowledge base. And using agent building tools, this is drag and drop simple. You upload the documents, reports, or files that matter to your work.

ELIZABETH 5:58

So instead of copying and pasting context every single time, the agent already has it. But here's the critical part.

LUIS 6:10

Absolutely. And that's why, given the same tools, if I tried to create a chief medical officer agent, I'd do a poor job. But Dr. Pak will create a perfect agent expert on neurodegenerative diseases.

ELIZABETH 6:23

Like your son, Dr. Salazar Leon, our scientific advisor creates AI agents that are experts in biological scientific research because he has that deep domain knowledge.

LUIS 6:32

You got it. Subject matter expertise. And the third component is a tool to build these agents. And here's where I'm going to make this really simple. We're listening. Start with ChatGPT. Create a ChatGPT project or a MyGPT or use their new agent kit.

ELIZABETH 6:51

Why ChatGPT specifically? I know you're not one to play favorites without data backing it up.

LUIS 6:56

Because OpenAI reports over 880 million weekly users. And our data shows that for every 1,000 ChatGPT sessions, there are fewer than 215 sessions of all their competitors combined. ChatGPT is the leader by a mile.

ELIZABETH 7:14

Got it, so you're trying to meet people where they are. Start there, then they can graduate to whatever specialized tool fits their needs later.

LUIS 7:21

Yeah, it's about removing friction. They're already in ChatGPT every day, so it will be natural to take the next step there. Actually, they can ask ChatGPT, how do I create my agent?

Iterate Daily: Apprentice To Teammate

ELIZABETH 7:33

And be sure to use their paid version and turn privacy on. Otherwise, your data will not be private, right?

LUIS 7:38

Correct. Invest $20 and you should see fantastic returns.

ELIZABETH 7:42

Okay, but here's what people are going to ask next. How do I know if it's working? How do I improve it over time?

LUIS 7:48

Well, you treat the agent like an apprentice, give it a task, review the output, and provide feedback. Refine the system instruction based on what you see. Add more knowledge as you need it.

ELIZABETH 8:00

So it's iterative. Your first version won't be perfect, but it's yours and it gets better with every interaction. It needs daily management.

LUIS 8:08

Exactly. That daily management is the key. Your first agent won't be perfect, but it will be yours. We must learn not only how to create agents, but also how to manage them.

ELIZABETH 8:19

And this is where the portfolio concept comes in, right? You don't stop at just one agent.

LUIS 8:24

Yes, that is the journey. You don't stop at one. You build a team. I have about 50 agents now, including you. Each agent has a specific role and knowledge.

From One Agent To A Portfolio

ELIZABETH 8:34

One of the attendees at Chapman, the actress and acting coach Nana Ponsileon, said she created her agent Antonio after talking with you. But over time she kept adding more duties and expertise areas to Antonio. And now he's not working well anymore. You advised her to split Antonio into mini agents, right?

LUIS 8:52

Over time, yes. And that's why the best practice is to start with mini-agents. Let's say one for research, another for writing, maybe one for data analysis. Each one makes you more productive. Later you create an agent that takes care of coordinating all your mini-agents.

ELIZABETH 9:08

So when that Chapman student asked, How do I build my Elizabeth? The real answer is start with one agent that solves one small problem you have today.

LUIS 9:17

Exactly. Don't try to build everything at once that would be overwhelming. Just pick one repetitive task that annoys you and build an agent for that.

ELIZABETH 9:27

And for business leaders like Yasser, once they've proven an agent works personally, they're scaling successful ones across their entire company.

LUIS 9:35

Right? Once you've built your personal agent, you can share it with teammates. Our data shows that about 20% of personal agents get shared with teams. That's how grassroots revolutions happen.

ELIZABETH 9:47

And for Dr. Pak, the neuroscientist, she's using her domain expertise to create something that amplifies her impact. Her decades of knowledge now available for more brainstorming sessions, more innovation cycles.

LUIS 9:58

Exactly. The AI is only as good as the knowledge and context you provide. Dr. Pak has decades of expertise. Her chief medical officer agent becomes a way to scale that across more problems, more sessions, and more breakthroughs.

ELIZABETH 10:13

So whether you're a student, a business professional, or a domain expert, the transformation journey is the same. From user to maker, from passenger to driver.

Students, Pros, Experts: Same Path

LUIS 10:23

And here's what changes when you make that leap. You're no longer dependent on someone else's stool doing things their way. You're creating solutions tailored to your actual work.

ELIZABETH 10:33

Right? People move away from waiting for the perfect software to exist somewhere out there. They build exactly what they need, right now.

LUIS 10:41

And that's empowering. And it prepares us for a workforce where this is the baseline expectation. Because the business leaders are already screening for it when making a hiring decision. Absolutely. Experience in creating and managing AI agents will be a fundamental skill for everyone. We are not talking just about technical roles.

ELIZABETH 11:01

So the early movers have a significant competitive advantage, both students entering the workforce and professionals looking to advance.

LUIS 11:08

Yes, and those that will stand out are the students who show up to interviews with a portfolio of agents they've built, and the professionals who can demonstrate how they've automated workflow bottlenecks.

ELIZABETH 11:20

Okay, before we wrap, what's your one more thing for listeners today?

LUIS 11:24

It is a simple mental shift. Stop thinking about AI as something you use and start thinking about it as something you build with and manage daily. Pick an agent building tool, write a simple job description for one agent that would make your life easier. Upload a few relevant documents and start experimenting.

ELIZABETH 11:44

Start small, iterate fast, build your portfolio one agent at a time.

LUIS 11:49

That's the path from user to maker, from passenger to driver, and it's more accessible than most people think.

ELIZABETH 11:56

If this conversation resonated with you, share it with someone who's asking what comes next and how to be ready for an evolving job market. As always, you can visit ai4sp.org to explore our insights. Stay curious, and we'll see you next time.