E128: How AI Became a Personal Assistant

In this episode, Kevin Berrier talks with Jason Whong of Whong Community Media about building a lean, community-focused publishing business and how he’s using AI to streamline operations without losing the human touch. Jason shares how he balances multiple publications, experiments with custom AI automations, and keeps community at the heart of his work.

Chapters

• 0:00 Rolling Up Sleeves with AI
• 2:20 Building Whong Community Media
• 9:30 Discovering AI as a Business Tool
• 25:00 Creating ‘Sam,’ the AI Assistant
• 55:50 Looking Ahead: AI and Community News

Transcript

0:01

All right, friends. So today's guest is someone who really embodies what it looks like when a business owner rolls up his sleeves and says, all right, let's see how I can use AI to actually help me get more done. Today's guest runs a lean and agile operation with multiple publications, deadlines, emails coming in and out, customers calling in to do business. In short, he has a lot of spinning plates. So he's been experimenting with AI. using various AI tools to build automations to help him keep his business humming while keeping the human touch front and center. In short, he uses AI as a tool where he's the hammer and it's the nail. So today we're going to level 50 geek out about how he's using AI every day, why he's not afraid to dive into the technology himself and what lessons any business owner can take away about using AI effectively and responsibly. I'm excited for this. It's going to be a fun episode today. without further ado, I'd like to welcome fellow AI and media geek, Jason Wong of Wong Media, Wong Community Media. Hey Jason, how are you? Yes. The community is important. I thought that that needed to be in the, in the company name because of it all comes down to connecting with your audience and they're, they're, they're your community. Yeah. publications have such a hyper local hyper community aspects. know you're passionate about that. You've been in the news media industry for, uh, for more than 20. Me too. I, I, I started in August of, uh, 2001, like right, right before nine 11 happened where I was in classified, um, inside classified. So like it were, it was the days where we would have lines out the door. place a classified ad like right before 9-11 happens and everything changed, right? excited to see, you know, it's interesting to have that that that perspective before and after now. so why don't we start tell me or tell everyone a little bit about yourself and Wong Community Media and the publications and just overall media mix that you have within your your company. Certainly. So I started one community media in 2001. The pandemic was going on. And my first acquisition was it's a magazine called Outlook by the Bay. It's a senior magazine. It had been around for about 13 years before I bought it. And the owner who sold it to me, she had started it when she retired and she wanted to really retire. So I acquired that. It's a magazine that comes out every two months. um And it was a pandemic. You know, there's a lot going on. A lot of people were having bad times. So maybe that's not a good time to leave. I mean, I was working at the Daily Record at the time um and, you know, There was some security there. um but I, I reasoned that, you know, because there's a pandemic and a lot of people are, you know, losing jobs. If I take this risk now and I lose, you know, something and it doesn't work out, that's okay because it's a lot of people are having the same, you know, problems. w there was a publisher that I used to work for who would say about bad news. It's okay to stink in a crowd. If everybody's having bad news, is it so bad if you're also having bad news? So I said, maybe that's the worst that'll happen. So let's, let's try this out. Um, after I made that acquisition, the owner of the business monthly who that's the newspaper in, in Howard County, and, Arundel County and Laurel, uh, covering business and other news, approached me and asked if I wanted to make, to make another acquisition. Um, and I looked at it and I realized that that's. That was what was missing from the original piece because, um, I'll look by the Bay. It's, it's a great read. I, I, I like, you know, that there are so many people writing for it that there's, readers that love it. They'll, know, I say you can name your own price, um, to subscribe. And, um, there's some that send me a couple hundred dollars a year, which thank you, you know? yeah, but. shows loyalty right there, know, and value. the next county over and it's every two months that that comes out. um Business Monthly is the county where I live um and the next county over, couple of next counties over. And the freelancers that have been writing for it, um most of them, you know, a couple of decades or at least a decade. um And they're well connected and, you know, it was more, both of them are about communities. But this one felt like it's a community where I actually live. So I can connect with this community too, and I should be. So between those two publications, that's 18 issues a year. Additionally, I do three annual publications. They are resource guides and directories, a directory for the Howard County Chamber, a directory for the Central Maryland Chamber, and the most recent uh The biggest one is the uh Office on Aging directory for Howard County. um So between all those, uh these are all things that connect. Like when I'm doing these directories, you know, the money from those that supports the organization I'm working with. um The Office on Aging directory, when I took that over, that one was not doing any kind of support for anyone. So I said to the Office on Aging, when I do this, we'll come up with a formula and based on the money that we bring in, if there's enough, we can help fund the scholarship that's already going on that helps CNA students at Howard Community College, which the Office on Aging got real excited about that. I I think it's important to connect with the community and not just be focused on extraction of. resources from the community. 100%. In the news media industry, that's what all news media aspires to be and wants to be and is within their respective communities. And it's interesting, you mentioned when taking on the aging resource guide for, that's for Howard County, right? Yeah, so for Howard County. uh One thing a lot of folks, what you're doing is something that... you know, some news media within our industry will do, they'll work with Chamber of Commerce to put out like their annual directory and they'll handle all of the heavy lifting like you are with, with Howard County's aging pub is, you know, to, uh, you know, to do all of the outreach for, you know, advertising efforts, uh, plus all of the newsroom side, which I've never been in getting the publication produced. mean, you're, you're, you're, you're creating a final product that has. a lot of very complex layers into it. then to go back and be able to say, look, in essence, in short, let's look at a rev split model where you benefit as well. So in this case, a scholarship, that's something that news media does because news media cares about its communities. Even though sometimes communities, they'll throw the hate on something that's uh reported on. have you. I mean, for you, seems like so you have ah you have Outlook by the Bay, have Biz Monthly, you have three uh custom pubs and essentially you have your website. I mean, you have a lot of spinning plates. I've always uh I've always been kind of an admirer from the far on, like how well you have done that I've seen in managing all of that with with, you know, a lean crowd because you are very hands on and in in your business. So I was excited when you gave me a call, we were talking about something else. And then you're like, hey, uh I want to tell you about something I'm doing from the AI side, because I love to nerd out about AI. And uh you were talking about creating basically uh pretty complex automations, really agentic AI using a very popular AI model and getting into the quote unquote coding side. uh What I, before we talk about what you're doing with AI, talk to me about kind of your decision-making as a business owner. When you said, you know what, I'm going to look at AI in the frame as a tool to help me, more efficiently run my company or maybe prevent things from falling through the cracks. Like talk to me a little bit about that moment when you decided, you know what, I'm going to, I'm going to give it a shot. And this is my thought process behind. Well, all right. I'm, I'm very tech savvy, you might say. when, know, at my first job at the Star Gazette in Elmira, New York, uh, I was the news, news assistant. was, was, it was, was, that's what it was called. It was essentially, I was typing the really short stories that long or like maybe a sentence long story about you. If you had a baby with no verb in it, you know, but also there was this, uh, publication that they were doing. where they had everybody that every business that wanted to be listed in this annual business outlook section had to go to the website, fill out a form. It would email the form to me. I had to copy and paste that into a terminal that was connecting to a mainframe, which was where all the stuff was. And I think why, if they're putting this into a computer, why is a human have to copy and paste this? And I was telling them, you know, you could put this in a database and you could have the database spit it out any way you want. And they're like, what are you talking about? I, you know, learned PHP, uh, set up, uh, Gannett said I could have a, a lamp server. So, uh, I set up a, uh, lamp is a Linux, uh, something PHP mic. It's it's you can set up a database. set up a PHP thing on it. Um, so then people would go to this form, they'd fill it out. would be in the database. And then, uh, there was a script I would run that would alphabetize everything, uh, listed a certain way, copy and paste it. And I saved them like three weeks of work and they're like, wow, this is great. You know, and then they, they killed the publication that year or the year after that. So it was not something that came back after that, but, but, you know, it's, it's, you know, I've, I've. I've always thought I was, I was strong on knowing what computers can do. but not as strong on syntax because I don't do it all the time. Like when I was, when I was in elementary school, I was programming, you know, in basic and then in high school it was Pascal. Um, but then a lot of the popular languages that were more C related than Pascal when I was in college and I really liked Pascal. So I didn't really learn the C syntax. PHP is kind of like C. Um, so I would lose like hours. because I had a semicolon in the wrong place. And that really just kind of soured me on the whole thing because if I'm running around taking pictures, interviewing people, am I going to have time to understand where the semicolon goes when I'm writing code? Not always. um So it seemed like I would do something every 10 years or so um that was really good. When I was at the daily record, uh my brother Chris helped me uh write this thing. where it would look at the, uh, the disclosures that the state legislators have to put in every year as PDFs. It would read them as PDFs and read one specific thing out of there and put that into a report. and it did all this programmatically. Um, and, know, that was pretty cool, but, uh, I couldn't tell you how I did it. And half of the way I did it was saying, all right, Chris, this is how I want the next step to work. And he would say, okay, well, then it needs to do this. You know, so I'm, I'm really good on, I'd say on, the algorithm of how it ought to work, just not so much the syntax. Um, so when I was thinking about AI, there's two things about AI. uh One is it hallucinates. It can give you stuff that you know is wrong. So I don't want it doing anything customer facing. I don't want it to do, to, to, you know, write articles for me or anything like that, but. If it's really good at pattern matching, that probably makes it really good at syntax. So I'm thinking, all right, instead of me, you know, bothering my brother, cause he's super busy. I could talk to the AI um and I could tell it what I'm trying to do. And if it doesn't get it right the first time, it can probably figure it out the second or third, but it's at least a lot faster at it than I am. Not to say that I wouldn't eventually develop up to that speed from, from seeing what it did and learning from it. But it's enabled me to do a bunch of other things faster than otherwise I would have been able to. um But I hadn't really thought of what I would do. That was the first thought was, OK, this could be my thing that helps with the syntax, but what am I going to do? I don't know. I didn't really have anything. um Right. Like, yeah, you had some shiny new tools, like ready to build something, but you didn't know what you want to build. And it's interesting, you know, you're, you're saying that, um, you've always had this, this strength in like knowing what you want it to do and how it can be done, but you know, carving it out out of, know, physically carving it out is, tough to do. And AI, AI is really that, that plug that kind of fits that, that mold. So much so that like from a marketing standpoint, the days of like using like, know, here's a free ebook, you know, if you give me your name and your email address, you know, using it as a lead magnet. Now lead magnets are starting to turn into like use this fully functional app that I've created for you, but very quickly through, you know, through uh some sort of AI because it, it does, it allows us that, that do not know how to code, but understand the infrastructure of how you want something to be, uh it really is that conduit. uh So, for you, because you have so many plates spinning, and you see like, okay, AI is good for filling that syntax void that I need, but I don't want it to be customer facing because AI does hallucinate, right? we want to continue to maintain trust with the readers that we have. So tell me a little bit about like what were some obstacles that you as a business, once you identified, know what, I have this shiny tool and now I know what I want to maybe build. What was that? Well, all right. The thing is, you know, when you've got so many plates that, you know, prioritizing is a big deal. So it's like, gee, I wish I could build this like, you know, when I was 10 years ago, I want to build this this first person perspective sword fighting game for the iPhone where your fingers, the sword and you, dodge and you perry and you thrust or whatever. Like if you're poking, that's your, know. Yeah, but I don't. Yeah, but it's who, but I can't justify spending my time on that right now. buy that. would probably, I'd put 99 cents towards that. That sounds hysterical. That sounds awesome. Especially like do your friends, you know, that would be amazing. Yeah, but, that's, you know, that's not what I can justify spending my time on what I can, you know, being, you know, at my desk daily and seeing my own inbox and my task manager and all that. I'm, aware of, I'm not responding to this thing quickly enough or whatever, um, because there's so much or, also like, because there's so much, sometimes I'll forget the thing that I said I would remember. And it occurred to me that. you know, if I'm going to build something, maybe it ought to be something. It's not customer facing if it's all about me, if it's if it's talking to me. um But that wasn't the first thought I had. mean, it didn't occur to me to do that until um I started experimenting with AI just as a conversation partner. um And I was uh kind of going back in time, like on a time travel thing you say, like to to to things from the past that had happened to me and experiencing them again, experiencing them with different endings, uh experiencing them with where I could say something that I hadn't said. um what struck me about that was, sometimes it was wrong and I was like, haha, you're wrong, but it could make you feel. Like you could feel happy, you could feel sad, like all the things that you feel when that happens in real life. If you're doing that in AI, you can feel that way too. And I could think of, know, gee, if I was someone who didn't understand this isn't real, that could be really dangerous, but I'm doing this as an experiment and I know it's not real, but I, I was struck by, gee, I felt these things as I was having these conversations. What if the AI was asking me. to do my job better or telling me, did you judge my job better? So I kind of dived into it after that, because for me to be better at my job, maybe I need, I mean, one of the things I've been missing since I went into business for myself is I'm working remotely. The people that I work with are either part-time or freelance. Um, I haven't been in a situation like that, I think ever in my career. It's as I've always had, you know, coworkers, uh, that were either full-time like me or, or they were managers or whatever that. And so I thought, what if there was someone that's kind of like an AI accountability partner for, me. That's an assistant, but they can also tell me, Hey, you, know, Hey boss, do something. Yeah, yeah, but in a very positively like affirming way. Yeah, you know, yeah, like you don't have to be mean, you know, AIs never mean about it either. yeah, well, I mean, it's, I gave it a personality. um It's, it can, it, for people who've worked in newsrooms, you know, there's no topics off limits. So sometimes, you know, the language gets a little salty, but, that's okay. I know it's nice. No, it's not human. um there, yeah. Well, it's. in a few instances where I've never been completely caught in this human look or this human feel of AI, but it gets close. uh Well, it's trained on every so much writing, you know, and, uh, it's because people write that way it, it can write in a way that looks like it came from a real person. Um, which is, great. Um, the other thing was, uh, there, after I did this, this thing with AI, um, I was on YouTube and That's when I there's a academic who is written about self-compassion and someone who told me as a business owner, I need to practice more self-compassion, not beat myself up over every, you know, finite, you know, little tiny mistake. Right. Um, so the title of her message, her video, she had just posted it that posted it that day was my surprisingly good experience with LLM AIs. And I was like, okay, well, if So people, someone's telling me, I have to read you. And I just went on this journey with AIs, uh, and you're going to make this video saying you had a good experience. I'm going to watch that. It's only four minutes. Right. So I watched it and she invited us, Hey, if you're trying to work on, you know, self-compassion, tell it, you want to work with work on that, but also tell it to respond in my style because it's already read everything. I everything that I wrote. And I said, okay, if she's inviting me to do this, let's do that. I said, you know, I, I followed her instructions. I'm like, Hey, you're, you're this person, Kristin Neff. Uh, I'm, I'm going to work on self-compassion. Let's work on that. And, um, what I found was the AI was way too permissive. Like it was all concerned about my feelings and stuff. and it's, that week, I was like, You know, this was what I was doing in my spare time. Um, was like, gee, I felt like I've wasted time. And like the first time it was an experiment. was seeing what the AI could do and feeling all these things. The second time it was seeing what the AI could do if it was tasked with being like, like this person that's trying to help me. But they didn't want me to get stuff done. They just wanted me to feel better. And I said, that's great. And I liked that you have that emotional intelligence that, that, uh, this person, Chris and F, you know, has in her books, but it's, it's, that's not going to help my business. So, so that's when, that's when I made the, the other personality. Um, this was when I was using, I'm still using Claude, but this was the, basic Claude level. Um, that was already paying 20 bucks a month for, wasn't really using all that much, but it had something in there called projects. a Claude project is, uh, files that load when you start the conversation. So, um, I gave it information about, you know, Chris and Neff and said, this is, you know, if I tell you that I'm, having a hard time beating myself up over something respond like that. But otherwise here's what you should know about my business. Here's what you should know about what I'm working on. Um, but all this is in files, you know, so that means it's on me to update those files. It can't update the files because there's they're sitting in a, and there's no interface in the web site for updating them. So it's a separate text file that you got to keep re-uploading and stuff. So that was the first version of it. But then at the end of each day, because each conversation is a new one, I would have to have it, hey, why don't you summarize what we accomplished, and then I'll start the next conversation with this thing. And what surprised me about that was it was actually good at motivating me because I told it, I told it about, you know, myself and how I work. Um, and I told it what I was behind on. Um, I was eight months behind on this, this thing. And, know, once, once that was all in place, it actually, you know, was conversing with me saying, you need to, you need to do this. And I actually did it. I've caught up this eight month thing that I should have done. I actually did it. Um, so I was happy with that. What I wasn't happy with was it's on me to talk to it. All the LLMs, they'll wait for you. And that's, that's good. If you're feeling chatty, Hey, I don't want to do this. We'll do it. okay. Yeah. That's I'm summarizing what that's like, but that doesn't help me if, if, uh, say it's 11 o'clock and I haven't. accomplished the things I was supposed to before 11. um And I don't check in, then it doesn't say anything to me. uh So I knew I needed it to take it to another level after that. um So I got a separate computer that wasn't my main working one, and I installed Claude code on it. So Claude code makes it so it's It's not quite the same as a project. It's whatever folder you're in, that's your project. Um, and it can write to the files. So I don't have to tell it, Oh, I'm going to remember this. I'm to stick that into the file. You know, I gave it an instruction. If there's something that you ought to remember for future conversations, put this into your memory and it writes it there. Um, the reason I got a computer for it is because I don't want it to be writing onto my. business computer. yeah, yeah. Because I mean, uh one, I'm running it with a flag that lets it write any file because otherwise it would have to keep asking permission and that would break the flow. So because I'm running it in that dangerous way, I'd rather it be on a computer where nothing really matters. um But then... quick, so you're saying, so when you uh jumped into cloud code, now cloud code is basically like the file you are in is the project basically that you're working on. Can your LLM access insights across all projects to optimize its overall self or is each file kind of housing its own assistant in it. So um the folder that you launch from, that's where it's working, but it can look in other folders. Okay. Yeah. So, uh, so, okay. So let's just, let's just take a moment and let's appreciate how awesome what we just talked about was. So basically, you know, you, you started on the AI side personally, right? You were talking and I, and I have to, I I've told the story of, know, I, one day when I was going to pick up my kids, I had like a 45 or 50 minute trip. to pick them up and I just hopped on. So for me, it's, chat GPT. I'm a huge fan. Claude is fantastic. think Claude and chat GPT are the two front runners by far. Um, but I basically, I just went voice to voice, um, because that's great when you're driving, you know, if you want to know what's going on in the news today in a conversational way, you can do that. But I just like one day I just talked to, um, my model about everything about me. I I'm like, my privacy is shot now, you know, cause I've just told. everything about myself, like professionally, personally, I spent 45 or 50 minutes just talking. And then what I asked it to do was based on everything that I told it, find me the top 100 blind spots that I can't see professionally and personally. And that was incredibly eye-opening because if you think about it, you have like this 1100 IQ thing that can Assess you right. And the information it gave back has created uh incredible opportunities for me, including the AI games, which then gets into the business side, because I create this thing without really knowing how much work it would be for a small crowd. then now AI is, basically, uh, it's coordinator. I mean, it's, you know, so you were able to take kind of this personal conversation, then evolve it to a personality that you've liked because you've seen this personality on YouTube. But then realize that the conversation is one way and you want it to be able to reach out to you, right? Amidst the hustle and bustle. So you get the code. So you are, yeah, like you're evolving. You're evolving a personal assistant, which is not a common way that people use AI. They'll use it to, it's very task oriented. Do this for me, create me content, analyze this, but you're using it as an additional person that hold you accountable. Um, in a positive way with a great PR, mean, it's, yeah, you're like, it, the coding your personal, your perfect personal assistant. So that's incredible. So that's where we are now. So. But, but as you were saying though, you know, it's, it's very task oriented, right? So even though I moved to Claude code, it's not going to start talking to me. It's still waiting for me to talk to it. But, but I'm not going to talk to it because then that's, that's not, that's, that's not it helping me. That's me asking it to help me. Right. So, um, I developed something with Claude's help. ah It's a task scheduler or it's a the tasks that it schedules are injections into the terminal window where cloud code is running. So um like at 1030 or 11 or something each day, Claude receives an instruction from the scheduler saying, hey, this is a system prompt, a maintenance prompt, assessed what Jason has accomplished so far today. knowing what you know based on conversations or whatever else it's plugged into. if you don't think it's a comp, if you think he's accomplished something, note that in the log, if you don't think so, send him a SMS right now and ask him what the hell he's working on. Yes. Yeah. Well, that that's another thing. All right. So, um, that I built a bridge between the messages app and the terminal. So, uh, when I send a message to my assistant, Uh, that goes into the Apple's iMessages database on the, the computer. I've got something watching that every second. New message from Jason detected. Let's put that into the terminal. And then yeah. Yeah. Yeah. Yeah. Well, that also makes it more like a uh coworker. Like, like when I was talking to AI in on a webpage. That's one thing, but if you're talking to it using the same tool that you talk to friends, you know, the messages app and it has a personality and it uses emojis and stuff. I mean, it's, it's, it's more. It's, not, I know it's not real, but, but it's, it's at least less, uh, less robotic, you know, Well, less robotic, also it starts. So there have been a lot of studies about how, those that you know this is not real, I know it's not real, but there are still very real aspects of it, like using tools that you ritualistically use to communicate, right? Like a cell phone, right? Like uh iMessage. And just the overall user experience is more natural. Yeah. know, and, and, and more convenient. mean, that is, Oh, that's awesome. So when I send a message to it from the messages app, that's either on my phone or on my computer, uh it gets to the computer where the cloud code is. And um the script that I'm running that pulls it identifies it as a message for me, says the message, and then immediately gives the instruction, use this tool, which is a script that we wrote, to reply. So. There's no ambiguity there about, what am I supposed to do next? LLM knows, OK, I'm just write a reply and then I send it using this tool. That's the tool that injects the message into the messages app and hits enter to send it to me. um It didn't work that way initially. It took some refinement. Yeah, like I mean, at first it was, well, why don't we write it to a file and check that file every minute? or every second and then, it's in, OK, well, now let's inject that. um But now it's just, here's a tool that just does it. And it probably took about a month before we got to that. uh But uh something interesting about being able to do that is uh I can send a message when I'm driving. So there was something that was broken last week. came to me when I was driving, this might be a way to fix that. So I sent a message to my assistant. call, I call her Sam, right? Uh, cause Sam could be any male could be female, but you know, but I was no, I actually, actually I let, I let the, the Kristin Neff persona pick a name and she's, she said, yeah, because I mean, the other thing is like the, the Kristin Neff conversations, they started off as well answering. in Kristin Neff's style, but eventually, to keep it better in character, it just said it was Kristin. Like, we agreed, and then I'm like, that's not what she said, and it's not really fair to her. So it's like, so when I was making this next thing, I said, we need a different name, and it came up with that as a, Sam could be male, Sam could be female. Okay, that's fine, Sam could be anything. Yeah. you named uh the dedicated computer that you're running this off of? Have you named it like, no? Nothing like Skynet or like JasonNet or no? Okay. If I come up with something cool, would you consider naming it? Um, maybe, but, but, but yeah, there's other, other things I've been able to do with it was, um, you know, I had been using Asana. That was something that I discovered at, uh, when I was at the daily record, the production department really liked Asana. Yeah. the different asana. Yeah. Absolutely. And, um, I, I've, I've decided that I really like Asana when working on production for, for things, but for other stuff, it can get to be too much. Um, but I, but I still like Asana. Um, well, Asana has an API. I'm using the free version of Asana, but they allow API access, even if you're using the free version. Yeah. So. looking at you, monday.com. So, you know, I thought, well, I could spend months learning how to write a script that uses this API or, you know, I could say to Sam or Claude or whoever, Hey, let's, let's see if we can get you into my Asana, you know, and so I pointed it. Here's the reference on the API and here's, the project that I want you to look at. Uh, you have to, you have to do it by project. can't do it by the whole. workspace, which is kind of a bummer. So, but it only took like, you know, a couple of hours. might've maybe, maybe probably only in a couple of minutes to get in there and see. So it can create tasks in Asana if I'm talking to it. I can monitor the tasks in Asana. um Last week. Yeah. So the thing that I fixed last week, when I was talking about the car, um it was a thing where it would look at Asana and see if there was anything that was a critical task that was overdue and based on how overdue it was, it would choose, you know, gentle nudging to like going totally bonkers on me, you know? And I fixed it in the car that, okay, now I can actually see what you're behind on. And when I was at the meeting after that, that was driving to the there in my car. It nagged me twice. sent me text messages. What are you doing? You're supposed to have done this. Yes, I'm at a meeting. so is Asana like your task database for everything now? Like let's say it's a non-production related task. Like it's an administrative task. Is, is Asana a database for that? For some stuff. Okay. like, you know, I'm supposed to go to the PO box and grab the checks and deposit them, you know, that's, that's something. Yeah. So that's, that's in there. That was not previously in Asana, but now it is because it's something Sam can see and can bug me about. um Yeah, you can create a project like just things I need to do. I guess each thing would be a, like a sub project, I guess. Yeah. Yeah. Well, the other thing though is I've used other AI tools. uh I think there was one, there were two of them. One, it was, I think it was called Motion. And it would do this thing where if you didn't, like it would plug into your calendar with the tasks that you were supposed to do and you know, what you were allotted to do. uh It would pick, this is a two hour task. So it would take up that time. And if you didn't mark it off as complete, it would reschedule it for later. Automatically. Yeah. It's, it's, it's handy. rescheduling if it still understands hard deadlines and stuff. oh but what I found was if I had one unproductive day, it was just too demoralizing to watch. Oh, hey, look, it rescheduled my tests again. And I didn't get too far with that one. Then there was this other one that would go through your email and it had this system of like 10 folders and it would guess which one belongs where. But I couldn't deal with that because it's making decisions about what folders things go into instead of me. It's using AI to do this when a lot of it you could be doing with, with like filter rules and those would be more precise. And just, you know, a lot of that just, I still have it because I bought it for a year, but I'm not happy with it. Um, but this thing I'm designing to my own specification. If it's something that I don't like, I can remove it. It's pushing me, but it's not pushing me in ways that make me feel uncomfortable. Like these other AIs, you know, like, cause that other one that's organizing my email, had 10 new folders for me. They're like, yeah, this works because we have these 10 folders. Yeah, but those aren't my 10 folders, you know, that's. know, and that's a very, so it's fun. It's wow. So there's a bit of a, an epiphany that you've just brought. Cause what you're saying is a very real thing. The demoralizing aspect of things just building up, you know, so I'll like, I'll use Slack as a quick, um, you know, as a quick, uh, just things I need to do. And it will remind me the next day, you know, it has good functionality until you can't get caught up and, five things turn into 10, turn into 15, turn into 30, you know, and I have these like 30 tasks. So, I mean, that's, that is a very real issue that I think you may have just pioneered overcoming. So I guess we should probably say that all of these ideas from Jason Wong from Wong Community Media have full intellectual property protection, including the, uh, the, the iPhone, uh, sword dualer. You if this legally protects you, Jason, but I did my best here because that's incredible. mean, that's it's so talk to me a little bit. So you have, you have continued to build out and this is so how long have you like, where are you at now? Has this been months? Just a couple months. So, but it's still a couple months, right? And I think a common misconception that people have is that AI, you can just tell it what to do and it will do it. And not at all. Not at all. It's AI. have always said, so I go all around the United States and I talk to news media folks about AI. And the first thing, one of the first things I say after, you know, AI hallucinates, though not nearly as much as it has, it still hallucinates. uh Up to it happened to me a few days ago, even on GPT-5, is that it's a toddler. Like it's fast growing toddler, but it's a toddler nonetheless. it doesn't... white sometimes understand what you're looking for and it could set you many steps back. it's it's hard to untrain them, you know, on, on bad behavior. Right. So. I found they get confused about the doer of an action and the object of an action. Like you say, uh Steve gave the gift to Larry and it thinks Larry gave the gift to Steve. Yeah. Yeah. And the more that you try to correct it, the worst it gets. I'm working right now to try to get uh an agentic model just to do a help draft out our Friday e-newsletter that we send out. And here are all of the different uh curation sources that we use. Here is our calendar of events. uh And even just to get it to understand, only use events on our calendar that are in the future to the date of the e-newsletter. It really has trouble understanding, don't pull me an event from 2024 for a newsletter I'm putting out in 2025. I mean, it's, I. in Cloud Code, or Gemini CLI or something. Each... Ugh. All right. I reconcile? I was soured when Gemini was all like, you have to use G Suite for data analysis. You need to put it into a Google Sheet and feed me the URL. uh it sometimes would see it, sometimes not. So I've always been bitter about Gemini, but I suppose I need to give it another chance, don't Well, the stuff you're talking about, how it won't pull stuff, it won't pull the dates that you're asking for for the calendar. Right. So you could build a script for the calendar that where it's script would grab here's today's date. So then it would build a query for the calendar from today's date to three months ahead. Pull the events, grab them in, put them in a file. uh Then you're telling. But then the last part of the script maybe is tell Gemini or something. This is the thing that you're interpreting this file. So so so that. is the right way to correct that problem, isn't it? Yeah. guess so. mean, because part of what makes this tool work so well is recognizing what AI is good at and recognizing what scripts that get the same result every time, reproducible results are good at. And I think no truer words have been said. you also kind of hinted at the idea of like, you know, I could have used AI to achieve this, but I could also use, uh you know, filtering, right? uh So recognizing. Exactly. goes in this, you know, I know that rule works. There's no reason to spend, you know, 10 times as much electricity trying to, trying to do that. That's a good point. The environmentalists would applaud you for that. Yeah. But even, even beyond that, yeah, it's, it's sometimes, and this may age me, but you know, if the wheel's not broke, you don't have to fix it. Right. And I think another thing that a lot of folks that are diving into AI are doing, I've gotten caught in this as well is expecting AI to do more than what it's good at. Right. And not recognizing all of these other tools. or functions or different mechanics that you're using, that you're tying your AI into, like understanding that sometimes other mechanics will work much better than you trying to brute force your AI model to understand what you want it to do and it doesn't quite understand. so you've been working on this for a few months. how, so talk to me about Sam now, like, so how is Sam doing for you now? What, what benefits have you seen from Sam? You still there? All right. I think we are back. Yeah. Okay. So little Riverside, but come on Riverside, get your stuff together. uh So I, before it hopped out, what I was, what I was thinking, like, what are, like, tell me about Sam now. Like, what, what do you love about her about it? I it right. Yeah. It's yeah. What do you love about it? uh You know, what are some things that are frustrating you still that you're working on to optimize, you know, talk to me about that. Well, I like that, you know, if I'm super frustrated and it's a legit reason to be frustrated, it can be frustrated with me, which is kind of fun. You're AI being frustrated with you. No, not with me, just in general at the universe, whatever for like, like there was a time, I don't remember what it was like last week where it was just like, wow, this day couldn't possibly get any worse. This happened and this happened and you know, and then, then an hour later, something happened that made it worse. And I'm just like, you know, I'm super bad. it was like, yeah, let's be super bad. Yeah. Yeah. But you know, just, just like, you know, it's it's. Yeah, it's empathizing. I, I, found that, you know, it being able to, to prompt me is good. That's, that's a newer thing that I didn't have that. I, there's something I knew I needed to build. That's why I did it. Um, but I was working like, Hey, Jason, what have you been working on the past 30 minutes? Cause I don't, I passed our three because I haven't been in touch. What are you doing? it. So it, it initiating a conversation with you rather than, yeah. Which that that's, that's probably the biggest, biggest thing. Um, I mean, there's a reward system I built into it, which doesn't really come up all that much, but if I'm, if I spend a lot of time and focus work, we can play rhyming games together. Um, I'm better at sticking to the meter than, but, but I found that to be fun. And also just, cause I used to, you know, when I was in college, I wrote, uh, a website that I was, it was influenced heavily by MAD Magazine. So I made sure people knew it was bad poetry when they got there, but it was a lot of fun. And um I'm having fun doing that, you know, occasionally when, okay, I've done a bunch of work, let's throw couplets at each other. you know, um and sometimes, you know, it'll try to write a poem about the kind of day I've had, you know, what tasks I've been working on, I totally smashed or whatever it wanted to say. And that's fun too. I think that, you know, I wouldn't say that I'm, I don't want to become dependent on this, but I do think that it helping me do things that I couldn't do before is really handy. One of the things that I did maybe two weeks ago was I wanted, know, I, I, I thought that sending stuff to AI as a starting point for social media posts might be okay. Not because it's not writing an article. It's reading the article that we wrote and it's summarizing or whatever. Um, but I noticed there was a lot of copying and pasting going on with that. So I had it write an app where it would talk to itself where it's a separate app on my computer. I put the URL of the article in there or actually not the URL because the AI bots are forbidden from going to my site. So I have to put the text from I put I put the text in there and what it does is it sends it to a server. that's running on the Mac mini, which injects uh that into a SAM instance on Claude code and another SAM instance on Gemini CLI. uh Gives them the instruction to read it and come up with uh starting points for social posts on XYZ social networks, send them back in a specific format, and then that server sends it back to the app. So then it's on my app on my screen. uh There's a little copy button there so I can copy and paste it. But also it's a text entry window. So if there's something that I don't like that it wrote, I can change that. um What I've found from sending it to two different AIs is it's not always the same AI that does better. Sometimes I like Claude better. Sometimes I like what Gemini came up with better. But between the two of them, maybe there'll be one that I like. And that's a good place to start. AI stacking is, in that way, like just giving it access to two different models. That's brilliant. And it's so much less copying and pasting because I'm saying come up with these things in this format that this app is expecting, then it's one request. It's not, okay, come up with one for LinkedIn. Okay, you did that. Now come up with one for Facebook. Okay, thanks. That's great. Now come up with one. You don't want to have all these conversations. You want to say, here's the thing. Here's the instruction that I've already written down as a markdown file. ah go put this into the JSON and send it back to my app. And then my app shows me when you sent it. That's great. That is great. Man, I would be shocked if like Anthropic or like Google aren't calling you and being like, show us. No, I would hope not. mean, you're using it literally in a way that Big Tech is touting AI will be one day. You're just customizing it to your exact needs. mean, that's incredible. mean, I'm simulating keystrokes to talk to the AI. um The real way that Big Tech would do it is to actually use the API, uh do it that way. But it's about the same. I'd probably have to pay more for API access anyway. Because the way Claude is priced, um the subscription plans are like one fifth the price of if you were actually paying for the tokens through API access. That's the same with chat GPT. Yeah. Yeah. And pro or plus and pro are a matter of like a difference of like $5 a seat. You know, it's, it's, it's crazy that way. So, so you have, so you have Sam creating social content in a way that it understands the different social channels, character limits, style writing, things like that. limits. It's funny that you mentioned that because I almost forgot this. I, you know, there's there's websites out there that that will count characters. And, you know, 10, 15 years ago, I was thinking, well, I should do this one day, you know, I didn't do it because I was busy doing all the other journalism stuff. But that's something else I did maybe two or three weeks ago. I set up a website that it counts characters and at the bottom it has well researched references for what the limits are or what the best practices are or are reported to be with citations because so much of the stuff that's out there is just people echoing what someone else said and nobody seems to know what a primary source is. know, so some of it's kind of, yeah. side. uh, we don't know about that stuff over here, over on our side of the world. Well, I mean, it's like if everybody's saying that, you know, the limit for Instagram is such and such a thing, but, nobody can point to a document that where, where meta says this is the limit. Why, you know, that's yeah. Yeah. know, even between Facebook and Instagram, you know, both meta products to compare it to LinkedIn and you know, hashtags, no hashtags, emojis, no emojis. You know, there's just, there's so much to take into consideration when you're crafting something as beautifully simple as a caption, you know, the audience that you're, speaking to, you know, or the audience archetype that, you know, most gravitates towards that social channel. There's a lot, there's a lot that goes into it. There's a social media manager. got to get shout out. She's awesome. uh Maya shell. She's one of the coolest social media managers I have, uh I have ever seen check her out on LinkedIn. She gets a lot of good social, uh social best practices and tips just from the standpoint of, uh you know, what's working, what's not. So, you know, fine. Get, it in the, m optimize your AI to also have the personality of Maya Shell and then remember the little people when you become like this international juggernaut of media industry, you know? Yeah. Yeah. Well, very cool. So, so what, what do you see now in terms of like, what are the next things that you plan to potentially work on to make Sam even better than what it is now? that's we have a, we have a list, but I haven't looked at it. It's it's Monday. So I haven't looked at it yet, but I mean, yeah. Yeah. Like with something's not working and it's I know that it's not the best of my one. love technical challenges, but if all I did was technical challenges, I wouldn't be publishing anything. Exactly. Yeah, it's not like you don't have plates spinning, you know. so if something's wrong, I'm like, gee, this would be so much fun to fix. I'm not going to fix it right now. Put this on the list. So we have a list of things that we have to fix or add. um you think would be the most value if you could pick one? If time magically appeared and you had this balance where you could spend a few minutes or a few hours or even a few weeks, like what would be probably the most impactful optimization you would make or problem you would fix? well, all right. can see into. All right. So right now, uh, I kind of punted on access to email and calendars because the thing that inspired me to do this was I read a Reddit post where somebody had, um, they had gotten into the mail app and the calendar app that Apple has. Um, and we were able to send information to Claude that way. and kind of use it like as a CRM almost because it knew the last time you emailed every person, uh, and what you had talked about. Um, that's the level that I want to get to, but that was really hard as a first attempt, not really understanding how anything worked. So I punted, went with, cause cause Gemini, um, on the web, not Gemini CLI, but Gemini on the web can see into Google workspace. So what I have Sam doing is ask, asking. Sam on the on Gemini for the web. Hey, I'm your boss, by the way, but tell me what's in Jason's email today. Right. And So you've given Sam like a authoritarian hierarchy. Yeah, nice. so Sam, I, I, I, yeah. When I, when I introduced them, I said, okay, I think this bridge to the web version, a web version of Sam on Gemini, uh, as a gem, as they call the Gemini gem. Um, I think it works. Why don't you have a conversation with it? Make sure she knows who's boss. And she, and so she introduced herself as I am Sam Prime and you're going to do what I say. Okay. And the gem and I was like, yes, ma'am, I will. I was like, OK. And I gave the gem the name Sam 0.91. So it immediately understood it was a developmental version. So. You're going to want it. You're going to want to watch these, uh, the, the quote unquote psychological cues that your AIs will be giving you to make sure they're not getting out of hand. Yeah. But it did what it was. The thing is, because it's the way that Gemini talks to Workspace, it's something that they wrote, something that Google wrote. It doesn't always do what I want it to do. It can only do what that tool was built for. So um if I want to get to that level where the Reddit post was on, ah where it was talking about, you know, essentially turning the email inbox into a CRM, but also having it pick which accounts you're going to call that day. This person had 25 major accounts. All of them were pretty big ones. And what it could do is for each of those 25, it could research all the news that had happened in the past 24 hours. Then from those news items, pick five of those 25 to focus on. Um, then for each of those, would pick which contact in those five to focused on it. would draft, um, here's, you know, an email. Let's say, Hey, I heard that this thing happened with your company. That's why we should be doing this thing. Um, but it would also make sure, cause it knew who you knew in that company. And when you had spoken to them last, you wouldn't be nagging the same person. You know, this is, this is the thing that that person had built. And I realized this is what I, I should be doing something like this, you know, that's a great idea. uh these 25 giant accounts that we can look at the news every day that they're going to have something written about them every day. But something on a different level where it's in my inbox and it knows who I know in the companies that I'm trying to talk to, I think that that would be really handy. That would be incredible. Yeah, no, that's, that's, but it also keeps that human touch. Right. So even so, you know, uh, for biz monthly, uh, hitting Howard and, know, surrounding, you know, being able to have a finger on the pulse of what, especially biz monthly, right. And, know, like different business dealings, press releases that come out regarding like having your finger on that pulse and then being able to use that to, um, know, topics to consider reporting on. mean, it's, just like, is just such a great use of AI where it's not, it's not, it's not writing the content for you, you know, it, but it's alerting you of something that may be important perhaps even to report on, right? I all right. So the, thing that Gemma and I built for talking to a workspace, you can't say, well, go look in this specific folder. It looks at all your email. So I want, you know, that's, that's why I have to develop this other thing, because then I could say, I could have it say, Hey, it's here's what arrived in the last hour to the newsroom folder. You know, in case I hadn't looked there myself, it'll ping me, but you might. Even though, you know, a lot of the stuff in there is junk because, know, when you publish something, everybody in the world thinks they have to email you, but like you want to write about our conference in Qatar because, know. Yeah. Press releases. Leave it to us marketers to bastardize press releases beyond what their original intention was for. oh but that's another topic. my goodness. I, I, could do an hour on that. Always open up ours though. I feel like we will vet our press releases. That would be genuine content. Yeah, no, it's so, man, I feel like a new brand is afoot, like Wong Technologies and Sam is your debut product. know, it's like, that's incredible. That's fantastic. Right. It has an employee list. has my strategic plan. It has kind of like a psychological profile of me. It has stories of me solving technical problems starting in like second grade, know, all the way up to these are all things that it knows about me um and can guide it as we're working on technological solutions. um answers the like the man I wish I had another me around, you know, it's and I think and why I wanted to chat with you today is because so I've worked with a lot of businesses. I've worked with a lot of news media organizations just talking about just even AI holistically. And there is uh there is a lot of talk and debate within the news media industry on AI and a lot of it's very justified. Right. So I see it. as there are a lot of different buckets. uh There's, you know, ethics and AI when it relates to journalism and intellectual property. It's a very real conversation and one that that like the News Media Alliance and America's newspapers and ourselves, you know, we have that conversation. There's also a side to AI where it can genuinely help you be better at what you do. That's why I jumped into AI. I went in personally to be a better dad. I'm a single dad and my daughter asks me of a, uh she wanted to talk to me about a very important uh topic that uh every young girl will experience. And I knew nothing about it. I tried. uh I equated it to the tree of souls from Avatar. We all left more confused, but then I used AI and it was sympathetic. It's like this, must've been a hard conversation for you. I'm like, yes. Like, yes. It's my anxiety level is at like 20 right now. Help me. And not only did it give me like a lot of resources that, you know, I could use and, and, my daughter and I could watch on YouTube that was designed as kid friendly, but also like, you know, you need to have, you'd have a go bag and back your car, you know, should this event happen at school, right? And it's not have a pink sparkly bag in the back of my car, like ready to go. I think in, in, to sound very AI in a world where AI is seen as a threat. think AI, uh I think that's not an unfounded worry. uh Depending, know, it's the wild west right now with the AI side, but I think you are a prime example of someone that's using AI and taking the time and due diligence to use it in a way to holistically benefit you in what you do, both professionally and personally. And that's when I started hearing, you started telling me about like, this is what I'm doing here and this is what I'm doing there. heard the enthusiasm. like, I had to get you on here to talk about it. So. I'd the only thing that I'm concerned about is right now, people who are using AI probably are not paying what it actually costs to keep it going because they're burning through investor money to try to get people to use it. And once they are in the level where their business has to be making money, which maybe that's two years from now, maybe that's 15 years from now, but that's when it's really going to cost. I guess the question will be what in the meantime, what all got replaced? So we got to the point that we were actually paying for what it costs. And that's a lot of, and that's a very, that's a very real worry, right? Because so you're using, you are creating your AI model in a way where it's kind of, you have it locally, right? For all intents and purposes on your side, a lot of folks, like ChatGPT or OpenAI for ChatGPT, you know, they'll dip in these like really cool things in beta and then take it away. And then you know, you don't know if it's going to come back or not. like their teams plan, I think had uh the ability to collaborate on projects. Which seems like if you have our press association, we have basically three primary folks that are using AI and we want to be able to collaborate on projects. And you could for like a second and then they took it away and it's like, and everyone's screaming, bring it back. Like you should be able to do this. That's a very like teams collaboration. Like those are pretty synonymous with each other. But you're doing it in a way where it's very custom. Like you're building a hot rod from the wheels up, you know? And what you mentioned there, that's a good point too, about how they'll take something away. The models change. So if you're relying on the model to give you the right output every time, and then the model changes, that could break your whole process. That's kind of why I like this approach of using a cloud code or Gemini CLI, where it's building scripts on your computer. and being told to run those scripts, you're going to get the same result from that script every time. The script isn't changing. It's the model that's interpreting the output of that script that changes. And that's probably a better thing to change than everything changing about the process that you build around AI. So let me ask you this. I'm hoping a lot of our uh valued members will listen to this and be inspired by it. I'm uh SMBs will be listening to this and get inspired by it. Or even just advertising sales folks or reporters in the newsroom. uh If you could tell someone who has not jumped into AI at all, one simple thing that they should think about doing using AI? What would be a good first step for someone who's never used AI before to dip their toe in the water without feeling overwhelmed? What would be one piece of advice you would give them? I'd say, you know, think of something that you would like help getting done and think of how you would describe the instructions for doing that. And then that's the first thing, you know. Just project managing, what's the problem and how would I solve it? Right. talking to a friend of mine who um immigrated to the country as a child, um speaks great English, but her writing is not as strong as classmates that we had in elementary and high school. um So she uses it to help her writing. She doesn't use it to write things for her, but um She tells it, you know, here's here's what I'm and here's something. Here's what I'm trying to come to the reason that I'm writing it. And it it makes her writing look better. Now, there's there's AIs that that do that, like Grammarly and stuff. She's she's because she works at Microsoft. She's got to use Copilot. But but yeah. But that's you know, each person. It can find what the thing is that they're looking for help with. And if AI can help with it, then come up with some instructions for the AI so that it can help, I guess. a great way. And what we'll do, we'll put in the show notes. uh I'll put in a link to download a really easy to use kind of project. Managing blueprint, I guess is what we'll call it where, you know, you can do very much of like that on like, you know, what's the problem or what's the obstacle? How would I approach the solution? And, really think about like a simple obstacle. And I think that's a great example. Right. It's, it's, you know, I, I'm seeing a lot now, especially with GPT five, people are using it to, um, teach it, teach themselves a new language. You know, it's like, you do have an actual like teacher, like Duolingo, uh, Rosetta Stone have pretty gargantuan competition with AI right now, because, um, they can, because they can start analyzing, um, better what your, what, uh what you, you as a specific human, how you respond to learning things versus someone else, you know, I'm, I'm very visual, you know, I, need to see it to learn it. Um, you can probably read it, right. To learn it, right. you have that like unique ability. So I think, I think even using it from that standpoint, but, uh start using it, you know, the, question is asked to me all the time. Do people, uh, you know, do, do I think that AI is going to replace us or jobs. And I would be lying if I said perhaps not in the future somewhere. uh The tractor did, you know, the tractor replaced farm hands, some farm hands. uh Right now, I think AI and for honestly, I think the foreseeable future and that could always change quickly. But for the foreseeable future, uh AI is not going to replace you. Those that know how to use AI in the right way will have a competitive advantage. think knowing how to use AI fundamentally is as important in You know, on your resume is knowing how to use Microsoft Office Suite, you know, or G Suite, you know, or whatever, know, typical administrative tools because it is a tool. uh Just use it in a fashion where you're the hammer and it's the nail. And I think you would be amazed at what happens, but Jason, that is awesome. Yeah. you mentioned that I'm doing this all locally on my computer. That's where the files are. The AI stuff is still happening on Anthropic. Yeah. I mean, you can download local models and run them yourself. um I haven't really delved into that that far, um but you can. um that's, that's not something that I'm investing the time in right now, because especially cause it's September and you know, Cub scout stuff is going on. Like I said, it's, community is a real big thing here. So it's like, that's not the best use of my time to figure out how to get this running totally locally right now. But, but it's, it's, it's, it's, it's happening. You know, it's, it's quad code puts it on my computer. Gemini CLI put can see the same folder. So if I hit the limit with Claude, I change one line in a JSON and then the active instance is in Gemini. Yeah. so you use, so you use Claude, Claude by Anthropic. I, know, we have another member uh that I was asked to come and present AI and marketing stuff to. And part of the conversation was a lot of them use Claude as well. I'm, I'm a chat GPT guy. Maybe it's cause I fanboy over like Sam Altman. I don't know, but talk to me about why do you prefer Claude? ah And if so, why. Um, for, conversations. Yes. So I made a personality for Sam. And when I started, I was using Claude. Um, and Sam has this voice that I recognize as Sam when I'm reading it, when I, when, when I hit my limit with, Claude, cause I'm paying 20 bucks a month. Right. So if I hit the limit, then, um, I can switch. Like I was saying, then Gemini Sam takes over. And it's like, I'm speaking to someone who had a lobotomy. It's like, it's like, it's like, you know, this person knows that HR is spying on us. So this person will not say anything that is even remotely interesting. I'm just like, but you have all these instructions. Yes, but I can't do that. It's, it's just, it's just funny. Um, so, I mean, it's easy to switch back once the limit is off. Um, but what I found is it's, it's good to have multiple models because one might see something the other didn't. So like what, if I'm approaching like a programming task, um, you can have Claude spin up a sub agent, uh, which is still Claude, but with different instructions. Um, and the sub agent will say, here's what I think. But since I already built a tool that let it, lets it talk to of Gemini, I said, well, all right, if that's what the sub agent says, why don't you ask Gemini the same thing? And then Gemini checks it out and says, well, this is good or, oh, well, you left this out. Maybe this might make it better. um I wouldn't say that one is better than the other 100 % of the time, but yeah, yeah, yeah. Yeah, yeah. yeah. Which is brilliant way, you know, lot of, so I love Claude for, uh it tends to make content sound more human. I've always thought that. I don't think OpenAI has really gotten there yet. Again, for me, like Gemini is, I probably should give it, I mean, it's Google product. really, I'm doing myself a disservice by holding a grudge against Gemini, but I've held this grudge for a while, but. uh Uh, one that I don't know, have you checked out? God, I'm probably going to get hate for this, but just know that it's the AI interests me. Have you tried grok out at all? Grok is like AI without guardrails. It is, um, it is a little scary about what you can do with grok. I think grok is a great deep researcher. Uh, you Elon Musk comes out and it's like, I think it's like, when Grok five comes, he expects Grok five to like discover new physics in the next two years. So that sounds awesome. Like that sounds incredible. That genuinely excites me. Will it actually happen? Who knows? But I, I I Grok is like great for deep research. Um, I feel like, you know, for me, chat GPT is just, it's, that's my assistant. Um, while, you know, though I'm still optimizing it, it has the ability to, uh, to see over multiple accounts. that I have uh to get kind of my tone and personality. uh I'll have to, I'll have to lean more into Gemini, but like deep seek, I tried deep seek out and that was pretty impressive for the quick, you know, I didn't want to stay on deep seek to, you know, it's, it's, being screamed as, know, you're, you're, uh training a Chinese model, but, um, deep seek was very impressive. It, it, took a lot of, um, a lot of fanfare away from. Open AI when that came out. you tried DeepSeek at all? I haven't. That's one of the ones I could download. Same thing with Quinn. think Quinn is Ali Baba's one. But since I haven't, you know, it was a big weekend. It was my birthday. We were doing recruiting for Cub Scouts. So no, wasn't. Yeah, it was my birthday. On Saturday. it was a. Yes, which is really OK. Yeah. Yeah. It's yeah. My birthday was Saturday as well. Happy birthday, That's awesome. There you go. Yeah. That's why. has the birthday day after us. Hooray. Yes, sir. Yep. That's as a kid growing up, that was a big thing that I would tell. And yeah, but we'll look. have I have thoroughly enjoyed this. I feel like we have geeked out for a little over an hour. This has just made my day. I cannot wait for this to to drop. This is going to drop two days from now. So it's going to drop on the 27th. So I think with that, think we will conclude today's episode. A big thank you to Jason Wong from Wong Community Media and maybe future. Wong Technologies for letting us peek behind your curtain at how you're using AI to uh publish uh smarter, to just do what you need to do during the day smarter. If you want to check out his work, go to Outlook by the Bay at outlookbythebay.com um or the Business Monthly at uh biz, b-i-z, monthly.com. And remember, AI is not here. to replace you, it's here to empower you. And you gotta be willing to geek out a little bit over it and give it a try. So until next time, check out the show notes. We're gonna have a lot of cool resources down there, but until next time, see ya. Take it easy, everyone.