Spokane native Paul Takisaki is pursuing a career in artificial intelligence. It's not his initial career field.
This interview has been lightly edited for length and clarity.
Paul Takisaki: I left because I went to work for Verizon and started at the front lines selling in a mall kiosk in Northtown Mall and thinking it was a temporary job, I was in between building websites before that and that just sucked me in. The technology, I am a geek at heart, and it was progressing so fast, I couldn't leave. And 14 jobs later, again, I never imagined being at Verizon that long, I had reached associate vice president, reporting to the chief revenue officer of business.
DN: By that time, he was 20 years in with Verizon.
PT: And it was at that point, we were about a year into AI, ChatGPT launching, and I was really distracted at work because AI was starting to take my attention. It was so exciting and I started playing more and more with that.
DN: Takisaki and Verizon parted ways.
PT: So I went from sort of being at the top of my game to voluntarily sitting in front of a terminal trying to figure out how to code.
DN: What was it about AI that so enraptured you?
PT: I started to realize very quickly, every institution will be affected by this and it will be the biggest consulting opportunity that has ever existed. That was just my feeling on it.
Now, I felt a little bit like we were cavemen playing with fire right next to our village and maybe we didn't know what we were doing, but it's matured a lot since then. I think that a lot of those fears have gone away and I just thought, hey, if I can learn this stuff early, I can be an asset to businesses and help them navigate the waters.
DN: So did you have some idea as to how those businesses would be affected?
PT: I did. At first, my gut reaction was, this is gonna eliminate a lot of jobs and they're gonna need to figure out how to use technology instead of that. My mind has changed on that.
I've been in business sales for 20 years and I know that business owners are strapped for time, strapped for resources. They're starving for information if someone can give them real information and I just saw this as an opportunity to do that for small business owners. That's more where my heart's at, in small business and how do you get an edge on your competitors? It's capitalizing on moments like this.
DN: So tell me how you learned. You talked about two years. What did you do in those two years to learn about AI?
PT: I was working about 16 hours a day. My first question to Claude back then was, how do I know I have Python installed? Like, what is Python? How do I know if it's installed?
And I built a gaming website, actually. I was a bit of a gamer. I haven't had time in years, but I built it and it started getting some traffic. And I was like, okay, this is kind of fun and I grew that up to tens of thousands of monthly visitors.
I was like, okay, I know how to do this a little bit. But back then the tools were a lot more difficult to work with than they are now. They've matured.
I started working on my first big product. It took me a year to build this out. Looking back, if I built it today, I could build it in a week. That's how far we've progressed in the coding tools. So that's how I learned, just sitting down in front of the terminal. ChatGPT and Claude were my best friends. And YouTube. It started slow, but then it started to accelerate.
When autonomous agents started becoming a bigger topic, I couldn't think of a single reason to use them. I'm sitting there like, what would you do 24-7? Fast forward to today, I have 30 agents running 24-7 and I could think of 20 more reasons right now why I'd want to spin up more agents.
DN: When you talk about agents, tell me what an agent is.
PT: Really, it is just a set of instructions for an AI LLM, or large language model, to follow the set of instructions on a repeated basis, and so you give it a personality, you give it rules, you give it permissions, you give it tools, things that it can do on your computer. Most people don't do this on their personal computer. They would do it on something a little less risky than that to get started, which I did in my case.
And then it has memory and so it has almost like its own brain. I think that's the big difference between going to a ChatGPT and asking a question, or having your own personal agent, is that over time, it starts to know you almost as well as you know yourself.
The other thing is they don't need to sleep and you can schedule jobs so that they are waking up on their own, running a job, shutting down, and doing that over and over again.
Now, a lot of the tools have advanced, like Cloud Desktop and ChatGPT Desktop App. They actually have that functionality built in. I really like to focus on the more practical side. I think that's far more practical than trying to run your own agents, just setting up scheduled jobs in there.
But that's how I learned, is just rolling up my sleeves. I don't have an engineering background. I was in sales, business sales, and I just took the leap of faith. I think that the lesson for others is that you can start anytime, because as I was mentioning before we went live, if I took two weeks off, I would be behind the curve. You can dive in now. The technology is advancing so fast. What I knew weeks ago doesn't apply anymore and I have found that a lot of this comes down to your imagination and I think the sooner you get into AI, the sooner you start to expand your imagination. Because today we can already 10X what we do. We just don't know what that looks like. We can't imagine that and that was the same thing with the agents. I couldn't imagine it. And so you start testing it and going, maybe, can you do this? Oh, you can. Can you do this? Oh, you can. And you realize those limits are much further out. It's us. We're the bottleneck in our imagination.
That's where I get really excited about this. You can almost speak something into existence in your life, a custom piece of software for yourself, without knowing how to code. It's pretty wild.
DN: So after two years, how would you consider yourself a different person from having looked into AI? Does it have that sort of effect on your life?
PT: You know, it has given me more time back to be with my family.
I worked, I had a national team. I had 500 people in my organization. I was working a lot of hours and didn't get enough time with the family.
And so now with having these agents, I'm able to 10X my work output. A lot of the jobs early on that I had to do, updating websites, that's all happening on its own, and not only is it happening, but then it has another agent check the work and another agent go and publish the work. So you almost start building these teams.
It's had a positive impact on the fact that I can be with my kids and my wife more and go on a vacation and not be checking my phone constantly knowing that things are getting taken care of.
So I think from that standpoint, it's a net positive. It can be a net negative as well, because I think the more you use it, the more you feel like you have a jet pack on and you can accomplish great things and that can be exciting and consuming because you start thinking of all these things and you're probably spending some time playing as much as you are working. I think there's a healthy balance you have to find there.
DN: You said you thought that AI was one of those things that was going to take people's jobs away and yet you've changed your mind about that. Can you talk a little bit about that? Why do you think that? Why do you think your mind has changed?
PT: I think it goes back to the imagination piece is that there's going to be businesses that don't have an imagination and they're going to say, I have these customer service reps handling 40 calls a month. I can give that to an agent and get the same work done for a lower cost. No imagination there. And yes, that can happen.
But then their competitor might have a bigger imagination and say, well, I'm going to implement AI. I'm going to keep the same headcount. They're not going to take all 40 of those calls though. We're going to use an agent to handle those. But my people are going to focus on those 15 customer calls that are at risk of leaving us and they're going to look at the data of all the calls and they're going to find patterns that we never would have been able to slow down and find. That's the company that's going to win in this scenario because now they're getting 20X productivity out of their employees. But you have to be able to imagine it. And I think the only way you can expand your imagination is to start asking it questions that you don't think it can do.
One thing I do before I go to bed ,and I would encourage other people to do this as well: I give it some big lofty goal. I want you to remake Photoshop for me overnight. I want you to test it visually. I want you to make sure that this looks like something that could be sold for a million dollars and I want you to run it fully autonomously. And I want to wake up and see a finished product. I'm blown away most mornings. I do it every night. I can't believe what it creates. I don't use half the things, but I just did it to test it and those are the things that I think expand your imagination.
So in your current job, you can start going, okay, I can see where I can show up and output this kind of work and be more valuable to my employer or as a business owner. I can alleviate some of the pain points my front line has in dealing with the recurring issues. We'll give that to an agent and let them do things that are more meaningful. So that's where my mind has changed as my imagination has expanded.
There needs to be a human in the loop always and I think that will always be the case. It will be less and less that we have to inspect, but there still needs to be humans involved there. So I'm very optimistic now. And I start to look at the past technology we've launched and think about how good that was and how many more jobs that created.
Sure, it's gonna be disruptive in some areas, absolutely. But that's any technology. And I think that more people, as they play with it, I would guess 80% of people aren't really using AI. I would guess that. But as more people do, I think they will feel the same way that I do and go, wait, I can do a better job in my current role and be even more relevant.
And look, if I'm hiring people, I'm gonna wanna know, are you a 20X person or are you a 1X person? Have you figured out how to multiply yet? And I think that's gonna become more and more the case.
DN: So we have a lot of societal angst right now about AI. You have elected officials talking about how should we regulate it? At what level? How should we do it? Do you have thoughts about that? What would be the most appropriate way to take this technology, use it for good, and cut off as much as you can the bad possibilities?
PT: Yeah, I think that you can't slow this down and we've seen just what China's doing with open source models to compete with our US big frontier models. So even if we tried to regulate and slow it down here, we can't stop the other countries from using those open source models and advancing theirs or China just accelerating ahead.
What I think we need to do is use AI to battle those issues and so one thing I've done personally, the phishing emails we get and phishing text messages we get, they're getting better and better, especially because of AI. I almost fell for my first one recently. I got on the phone with the guy and spent 10 minutes on the phone with him who said he was with Google and I started laughing at some point. I realized and he laughed too. He goes, was it really that good? Because he was using an AI script that was helping him respond to me.
So I personally have AI protecting me now against phishing. They stay in my email and they alert me. They actually put it in a folder that says suspicious and tag it. So I'm not even tempted. When I look at it for the first time in that folder, I'm already thinking this could be a scam. So you have to fight AI with AI. I think that's really the only way to do it. And that's gonna, I think, be a broad brushstroke across all of these issues we say AI may cause. The answer is not slow it down. It's just, I wish it was, but it can't be. So what do you do? You have the best tools as a country and as a business to protect yourself and protect your people.
DN: We finished the conversation by talking practical uses at home for AI.
PT: One of the things that you could think about is as a parent of kids, homework is a challenge. I hear every parent say this, and the way they do math now is not the way I learned to do math. I went to Claude and I just said, hey, can you build me a homework helper so my son can learn in a more interactive way? I just take a picture of his notes and it then turns that into video, audio overviews, a game, flashcards, a quiz, and then it has levels in here. So he can like achieve, you know, he's earning experience and he can achieve the next badge. He loves this thing. There was one night where I quizzed him on flashcards and he would have failed miserably. That's why I ended up creating this. By the next day, when his quiz was, he got 100% because it was more engaging.
Those are the kinds of things that I think a parent could be doing today when you go, I don't really know how I'd use AI. Do it to take pressure off of, from a homework standpoint. I mean, it has been a huge savior for me there.
I think another practical application is to connect AI to your email. That could be Claude, that could be ChatGPT, any of them. Connect it to your email and simply say, can you give me a daily brief every morning at 7 AM? And that's when you'll start to see, okay, this thing is an employee. It can work for me. It's a very simple task. It has no problem doing it.
The next step is go into whichever one you use, go into their connectors and just look through the list of their connectors and try to find a piece of software that you use in your life or you use at work and connect it to Claude or connect it to ChatGPT. There's many of them and it's expanding. All of a sudden you'll start to realize, oh, I use Photoshop. That's now inside of Claude. I have to go to Stripe to set up pricing. I can connect that into Claude. All of a sudden Claude starts connecting into all of these applications you used to have to learn and navigate. You never have to log into them again. It manages all of it from beginning to end.
So those are the kind of practical things that I would say to do and I think that that's how you start getting the imagination.
DN: So how would you suggest people determine which of the AI models might be best for what they're looking to do?
PT: That's a great question.
The only one that I would say certainly is if you're coding, I think that Claude has the advantage right now. But that changes by the week. If it comes down to any other task that you wanna do, like some of these examples I mentioned, it doesn't matter. It doesn't matter anymore. They are all so good they can do everything that I just laid out.
I think the best thing to do is if you can, try multiple, and the thing I encourage my friends and family to do is to spend 15 minutes a week in AI. That's not a lot to ask. On a Saturday, just play with it and try to test it a little bit. I would use Claude. I would use Chat GPT and then a little bit of Gemini because that's built into Google everywhere.
That would give you a good lay of the land because you'll start to realize they do have strengths and weaknesses. And I never rely on one as another tip I would give. I always use two agents.
So if it matters, I have one do the research and the other one do the checking. To give you an example, when I do coding, I use Claude and then Chat GPT checks Claude's work. 80% of the time, 79% of the time, it finds a bug. It finds an error Claude made. That's a pretty high percentage. And then it corrects it before it tells me that it's completed.
