Latest / Hardly Artifical Podcast / The Bots Are on the Org Chart

Transcript

The Bots Are on the Org Chart · Hardly Artifical Podcast

• Episode 8 final: Okay, Mike, I got a joke for you. I'm already nervous, but go ahead. What is a robot's favorite snack? I don't know. Computer chips. Wa wah. Man, that is I can I actually I'm a little ashamed I laughed there. All right, let's get into the episode. Let's do it. Alright everyone, welcome to episode eight of Hardly Artificial, a podcast where we talk about AI and technology, the latest and greatest, what's going on, as well as some real life examples of AI. I'm Derek Major. And I'm Mike Bukowitz. And yeah, episode eight. I mean it's also an interesting week too. Up here in Canada, of course, Remembrance Day is on Wednesday, as in down the States, it's Veterans Day as well. So, you know, Mike, it's I think we should acknowledge the men and women that are served b both our countries and h within our countries and also around the world too. Yeah, no, I agree with you. And it's it's definitely a d a week of I wouldn't say of sorrow or anything like that. I would say it's celebratory of the folks that that have, given their time and their lives towards, the greater good. So we appreciate it and yeah, you gotta you gotta you gotta know that we can't do what we do without the people that are defending it either, Totally. Yeah. And it's interesting story. So last week we were obviously in Calgary. On the way back we landed in Vancouver. those that you maybe not follow all the news up in Canada, there was an incident at the Wainwright military base in Alberta about a few weeks ago, and four people lost their lives. And on when we landed in Calgary, the captain came on and announced one of the gentlemen one of the soldiers that had passed away. He's from North Vancouver where I live. And his name's Aiden. And I noticed when we got on the plane that there was a man in a uniform and I also noticed that there was a couple other people. It it just something was going on for sure. But when we landed the captain came on and said that they had the privilege of transporting Aiden in his last his last journey back to Vancouver. And so after the captain came on and said that this is after we landed, I believe the ant got up and said a few words as well on our flight. extremely emotional. And and then they let them off the plane first. And then the really cool thing just really just shows you how much people care. Is on the way out when we're taking our Uber back home, I noticed there's a roadway that goes from Vancouver Airport, you know, up to a couple of bridges to kind of get you out to Richmond and towards Vancouver. there is a procession of ambulance, there was fire, there was police, and they're all lined up and ready for the motorcade to take Aiden to his his final resting place into the funeral home. it was pretty it's a it makes you know, as a Canadian, it made me really proud that, you know, that we really take these things seriously and we know we honor those, so yeah. Yeah. Yeah. Absolutely. Wow. So with that said, let's Jump into AI. I just wanted we wanna we wanted to recognize, it's special important week to recognize these things around, in our countries. And but of course we have AI going on too. And this week a few different topics. I know on my end I have a couple interesting ones. the first one I'm gonna talk about Grok, Grokbot. Mike knows all about me and Grok bot adventure. Of course, this stuff's always changing. I know a lot of you probably using Grokbot. It's really easy to use. I'm gonna talk about some examples there. The other thing I'm gonna talk about is this new scams going on, how people are impersonating different you know, CEOs and lawyers and to the tune of a hundred and twenty nine million dollars in scams, I'm gonna talk about some things you need to watch out for as some of the stuff gets really really you know, sophisticated. So yeah, I I mean Mike, what do you what do you got this week? I kinda lean a little bit more into the human side because I think With all the tech and all the hype that's going around and the fears, it just seems like we're ignoring some of the things that are actually happening in the background that is shifting our market and even the employment market to say of the reli you know, the the fact that humans within the AI world are starting to find their way back to the workforce, which is interesting. I I did not know this and I I want to explore that a little deeper today. The other one though, a little bit lighter side, but actually a little bit enlightening was Had an experience last week in Calgary that alluded to me researching a bit deeper of how many people actually know what an agent is and how to actually enable it. And I have some real data that matches the market data right off the bat with one experience. And I thought that was exciting to share because I feel like it's gonna be more inclusive related to where people are within this journey that they're trying to enable right now. So excited to share those findings, just like you are with Grok, specifically around personal experience. so next week I'm actually coaching an AI hackathon. And what's really interesting is yesterday we had a prep call. We're going through and so what the organizers have done, they've gone out and they actually asked a bunch of questions, what kind of problems people want to solve. And then there's like a a group of people that I'll be coaching and they talk they ask, you know, specifically what problems they're trying to address. really a lot of it's all very similar. a lot of people are using at least the group that I have, a lot of them are already using Claude and they're using it to build out a lot of things. But it's like they almost need that extra step to get them over into certain things. So looking forward to that. I'm gonna be talking a little bit about my AI setup, which we'll talk about too. But it's interesting as I start, you know, you start talking to non tech folks what they're doing, what things they're trying to accomplish. I and really the it's some of them are just really a chat bot stage as we've talked about before. And Now they're like, well, what else can I do? What can I do with an agent? And yeah, I that's why I get excited about on on the Grokbot stuff and the dot and the Muse and all these things that are really making it easy to set up agents for people. Right. Yeah, no, I agree. And that's that's one of the challenges right now. And I'll love to hear more about when you come back, kind of what what you found from there because the just the experiences alone and mine, which were actually technically tech very technical people, even showed the the elusiveness of what actually is an agent and what is it a c comp compile of. And that's that's a challenge in itself. So anyway, yeah, no, that'll be interesting. Well what what do you what do you what if you had to say in one sentence, like how would you define an agent? Ooh, that's a great question. I would say an agent is something that is going to enable a process that is not defined but is trying to reach a goal. And that that goal it can be has to be very simple. That's the thing that I think people don't realize is yes, you can go the route of, hey, I want one agent to handle my, I don't know, assistant or your executive or whatever it might be, but at the same time you're you're overcomplicating it already. And you're starting to stray away from what an agent actually is, which is basically having a employee, if I were to put it in standard terms, that has a defined we wanna call it a role in the company that can allow it to work within the parameters you set. But you have to simplify it enough because now you're straying if you're trying to do too many things, you're straying into a gentic, which is a whole nother level of complexity. So That's how I would d describe an agent because that's been the challenge for a lot of people to understand. But I don't know. What what are your thoughts on that? Well, that wasn't one sentence, by the way. Sorry. That's what happens. I had a long thought on that. it's been bottled up in me. you know, I I I I think for me it's Well y I I'm gonna do this I'm gonna I'm not gonna do it in one sentence either. I do like what Monday dot com's doing, how they're trying to they're putting a face to it. they have all these avatars that tie to it. But really that that avatar is like the face and behind it could be you could go to Replit and go build an app and call it an agent. 'Cause and it just happens to have a dashboard so you can see what it's doing. But it has a ser I guess here you go. One sentence. An agent is an AI Prompt that has a series of actions that can be recurrent. Mm. That's a good way to put it. Yeah. Yeah. So I'm probably probably expanding on it. We should probably get in debates with people on that one. But anyways, I I think I think it's an interesting agents is just like a cool thing, right? I think for sure right now. Yeah. well it's cool, but it's also the fact that people are not understanding the level of value they can get out of it because they don't even know how to start. I think that's gonna be our biggest challenge right now. So totally cool. All right, well hey, you wanna kick us off with Story this week? Yeah, sure. so I I actually was looking at one that I think we talked about a few weeks back and it was related to the you know, and it maybe but it might have been our first episode, I can't even remember. But the humans getting replaced by AI, you know, that whole fear around how AI can help companies downsize, be more efficient, things of that nature. You know I I'm gonna be honest with you, right now we're seeing the exact opposite. Companies are that replace people that they claimed was was because of AI. We know there's always another variable in that sense. are now one in five of those workers just in the US itself are coming back. They're getting called back because it's realizing that what we described as increase of volume of work is starting to become reality. And I think the the biggest idea and I I follow Propeller a lot because Propeller is consulting company within the management consulting firms and it ran more than one thousand workers in each of the US and UK between June and July and calls the pattern AI rebounding. so nearly one in five US are coming back. And in the UK the figure was thirteen percent of what was laid off is now either coming back or they're hiring for those new positions. So is this just the this is is this the about face where it's like I fired five thousand people and like oops I shouldn't have fired five thousand people. Exactly, exactly. And I that's where I wanted to go with this today is, it it's almost like a they said it's it's not only about the money, it's about the thirty percent said an apology or an acknowledgement from leadership, and six point six percent of the US workers said they would refuse to return under any circumstances because they don't they don't actually believe in what's going on now because it's like, well, why would we come back to something that could happen again? So I think the headline is misleading to a certain extent, but at the same time what I would say is It's telling of the uncertainty as well as the missing knowledge that is part of the AI world right now where people don't understand the power that AI can have, but also the fact that you don't know everything. And it's it's scary because you can't make educated decisions 'cause you think this is the direction and then six months later you're going, no, what did I do? Right? Yeah. So I don't know. What are your thoughts on that? Well, you know, I Immediately just makes me think about how I have all these AI agents doing work for me every single day. And man, like I feel like I'm do I I feel a little overwhelmed sometimes because I feel like there's so much work getting done and there's just more work to do. Like I feel like I'm taking on more work. And I think listeners think about all the things you procrastinated on over the years. Like so I'll get to this next week, I'll get to that next week. And I don't know about you, but my to do list is massive. And I suspect There these lists have gotten so big, but now it's like, hey, let's let's get all this stuff done. And I'm and I'll I'll go through in a second here, but I've I've got all kinds of agents doing all kinds of cool stuff and I'm getting more creative with how to specialize them, how to it's like the old days you go hire someone, you want them to be a jack of all trades or Jill of all trades. And man, they th they they got burnt out 'cause they're like swiveling, you know, hit on a swivel situation all the time. Whereas now it's like I have these specialized agents doing all these things. But the problem with that is they're they're all outputting stuff. It's like and you're like, holy cow, like so you can kind of do it a couple ways. You can kind of s you can, you know, phone it in for the day and but you got a lot of work done, just like you would have any any other time, or you can ramp it up all the way to the end of the day. And I feel like if you're in a competitive environment, yeah, you're probably gonna be doing a lot more work. But I don't I guess my point is there's not a shortage of work, and I think that's what businesses are finding out. And AI is just helping accelerate all that. But I think there will be a it's like that diminishing law of diminishing returns is now gonna it's gonna hit that way, whereas you can only get so much done. And I don't think a human in a loop's gonna go away because we still wanna at least I know you and I are like this. It's like we still want to deal with humans, right? So Yeah. Yeah. And I and I I agree with you on that. And it's interesting you bring that fact up of it gets a lot done. But you're the one that has to basically put it into action or whatever it might wherever it needs to go. So it's not it's not necessarily doing all the work. It's doing the administrative heavy work to get you to a point of reviewing what it did and being able then to produce value. I mean, I can tell you right now, there's there's some interesting scenarios happening and I see it in my own, you know, LinkedIn and, you know, email where like the slop is getting worse. Like it's just the communication is just becoming it's like a it's like I h how I would describe it is the differences between a local mom and pop shop e-commerce versus a you know, a rampant high production e-commerce that is on every platform. Like if you think about the difference between the two, you've got one that's really h focused h on high quality, customer service, things of that nature, they're trying to make a name for themselves. But those that have already I guess broken the ceiling when it comes to price price scourging and things of that nature, or maybe they've just been around so long they're too big to fail, that becomes more of a volume, you know, game, right? It's not not necessarily quality. And I I almost think about it as, you know, if we were to manage, you know, you know, we are we are managing people in terms of like what they actually do on a day to day basis, it's like, you know, write down what you do, but AI's never Well, it could. I'm not gonna say never, but right now it's not at a place where it can be creative and really I guess you could say value and keep that connection with humans. Like it's it's just not there yet. And it everybody notices too. Like everybody's getting fed up, burned out with the high volume of communication that's going on because if you read a lot of this stuff, it's just it's just generic. Like it's all the same. It's kinda like back I think it was maybe it was last week they were talking about how And and I I don't want to bring up anthropic day, that my goal, but I well I'm gonna bring it up anyway. is the fact that they were replicating video games. Right. And they had to shut that down, right? And it it's kind of the same manner is it learns something based upon something that exists. So if you're gonna create they were running they were running Nintendo Switch games on like iPad. Like Right. Right. Whoa. And and that's and they shut that down because they know there's a creative sense to that that just kind of it it just turns people off. So like I I guess to finish this one up in my eyes, I think, just making sure you take into account, and I we said this a couple of times to our team last week, was AI is not gonna be creative right now. You are the one that's controlling how it directs, where it goes, things of that nature. What we're doing is putting guardrails up for you to be able to work within them, but also not spend a whole lot of time on things that are just required of a business, right? Of documentation. backup information, updates, things of that nature. We're trying to take away that to focus more on that people connection. And I think that is something that again, going back to the basics, what is it that's taking away from accomplishing those connections and making them meaningful so it does feel more intimate in terms of the discussions you're having with the clients. Totally. No, I agree. Cool. Well speaking AI sloth, let's talk I'll I'll segue into Grockbot. So So yeah, that that's one of my two my first story this week is Grokbot. So I I'm sure many of you have heard of Grokbot, but and I've known about two, but it's it's like I found myself okay, I'll hold off, I'll hold off, just cause there's so many tools out there. Every week there's something new. in the last month we've seen Muse, Dots, and you know, Hermes as new releases and all that stuff too. But then I said, you know what, I'm gonna give Grokbot a shot. And what prompted this is you know, I I have all this extra admin work that's been you know left on my plate. As you know, as of last week I don't have my EA anymore. So so there's like a lot of stuff I'm like, okay, how do I like get smart about this and doing things I just never thought about doing and part part of two when you have an EA, an executive assistant, to forgive me for using acronyms, but when you have an executive assistant, it's How do you like you you don't know and you don't know? So you don't know how to ask them what things to do, but I thought, you know what, let's just throw this in there. And so I installed Grockbot. And I like almost instantly fell in love because of how simple and easy to use it is. It's it just for it it's a lot like the other tools, dots and all that, that do it, which I haven't tried yet. I know you guys at Oligia are looking at it. but it was like okay let's let's give this a go and let's see if this might be a fit for me. And it it became so seamless over the weekend I was able to configure it, give some prompts. And then what I ended up doing, I'm just gonna open up I'm not gonna share anything on on this, but I'm gonna pull up a screenshot that I took of my bots. And I I I went into my bots and I said, Okay, I can add the first one which is kits. Which is my my dog, my my dog that passed away a number of years ago, I decided, you're not gonna call my chief of staff Kitts. And so this is like the first bot. This is like the overarching orchestrating bot. And then what I started doing is I'm like, okay, well, what other activities would I like the chief of staff to do? So first of all, I hooked up my work email. I'm still against putting personal. I'm not I I don't want to have access to fine my personal financials and all that stuff. So I'm I'm kind of against that. But the work stuff I can control a lot more. There's service agreements as far as privacy and data goes. So you can hold people accountable that way. then I started building on my team. So here's my team I have. we have Alexandra. She's our podcast manager. I don't think you knew that. I haven't told you about her yet, but Alexandra. I see how it is. Okay. Got it. So she's she's taking care of the podcast and more about the marketing aspect, like how to promote it, what's working, what's not. I also have James, who's the webmaster for my major technology advisors website. So he's taking that over. But he's also getting direction from Danny. Danny is the SEO in and AEO expert. So he's doing analysis, reviewing the site, reviewing Google Webmaster Tools, just checking the data on that. Then I also have Ryan. Ryan is the lead gen. So he's actually scouring the internet looking for great opportunities for me to connect. And I'll give you an example. He He found a few people that were a perfect example to connect with on LinkedIn and crafted me a message I could use. And I just took that message and I tweaked it and adjusted like I always do with that stuff. And I go and I post it off to that person and I try to connect with them. We'll see how it goes. This is just like the last couple days. and the last one is actually Emma. And or actually this tube. Emma is my content engine expert. So she creates the content. So she actually takes the transcripts from our podcast. And she creates all the content for me to review. So again, AI slop's a big issue. We're just talking about that, right? I still don't like AI slop, but I like to use it properly when I'm using AI. So what I've done, I've created a good solid voice document. I had it analyze my tone, my voice, and how I speak. So it can actually create the content. I'm still adjusting and tweaking all of the posts and stuff. It's not like they just randomly go out. It's like, that looks good and It's not like it's it's good enough. It's like I'm actually reviewing and making sure it's working and it's sending it out once I've approved it and reviewed it. So it's helping me stay on top of those things. So all the messages this week on LinkedIn and Twitter and Blue Sky or sorry, X and Blue Sky and Threads have come from there. so I've really leaned into that. But then the other one that's really helping me, and last week we talked about hiring on Indeed using Chat GPT, right? So I also created another bot. I didn't give it a cool name. It's just called CS Expert Hiring. I think actually the bot gave it that name. But it's doing it's doing all the analysis from my email. So as people are accepting and booking the screening meetings with me, it's taking that and throwing it into the Monday.com board that ChatGPT's already been filling out. So here we go. I got multiple models doing work now, right? ChatGPT is on the Mac Mini. Collecting the resumes, posting them into Monday after doing a pre-screen. Now, what's happening with the expert hiring bot? It's going in, it's taking a look at my emails. Like if people, it's making comments if someone replies back after booking the meeting saying, thank you for the interview. It's putting that back into Monday. It's also taking the transcripts after I've done the meeting, the screening call with the candidates. It's doing a bit of a summary and dropping that into the Monday board. And also gauging it against the job post and the job description, also core values of the company. So it's doing all this work. So I have all these agents doing all these different pieces of work now. but on top of that, kids. Kitz has given me a summary in the morning, and this is my morning yesterday and today. I get up, I get a whole summary of things that's going on, emails to action, and I just go voice command with with kits. And I just we just talk through it all. I say, okay. John Doe has emailed me. Let's go ahead and create a draft reply for him, letting him know some options for a meeting. And then we just rattle it down and then it gives me the drafts to review the email. I tweak the drafts, I hit send, and then I might say accept a meeting. I had a meeting this afternoon. I'm taking my mom to the airport, so I had to move a meeting. So I asked kids, can you just go move the meeting and check the calendar of the other person, see if it's okay? It went and did it all for me. So in the past I would had my EA go in and actually reach out to the person, schedule it on all that. Now it's just doing it on its own. So that's been my foray into Grokbot and it's been so easy to use. previously Hermes is something I've been using as well, which works a lot differently than Grokbot. Grotbot, you're using whatever, you know, the models Grokbot's using, right? Which I did believe is I think they're using Opus five five right now. And I think The nice thing with Hermes though is I can I can set up a bot with different models. So I could have a bot with an open source model, one with Claude, one you know what I mean? This one's all one. But I think for a personal use case like mine, perfect. Yeah, exactly. Exactly. Now, I've got a question for you, and it came to mind when you're explaining this. To us, it's intuitive to the way in which things get architected and or defined r related to the types of agents. But In your in your humble opinion, and I wanna I want you to be honest here, how long did that take you to formalize in terms of how you were gonna structure this? Was this an overnight thing? Was this a couple days? Like just just let's give the audience a kind of a perspective around what it takes to get there in terms of an operational work orchestration just like you did. I can't imagine it took me more than an hour. Okay. And and here's the kicker. On my phone. Yep. I just I just I I had a con so I set up kits first and I had the conversation. And then in my head trust me, there's like tons of bots rattling around in my head that I still need to create. But this it really there's just kind of the heavy the the things I was thinking about the things I've been neglecting, like the website, I've been neglecting that, right? and now the webmaster has here's the cool the other cool thing too. once you start connecting things, so I connected Todoist, connected Gmail, my calendar, Google Drive. and Google, of course, go like Google Webmaster Tools and WordPress. So what I've done for the Webmaster is I have on to doist I have project in there for the webmaster. And I can put tasks in there, add this to the website, add that to the website, da da da. And it's just a Kanban view. And that's another way I've kind of tied it together. but I didn't realize it until I started connecting things. So I'm doing this all on my phone. I'm like, hey, this you what? I use a task manager. How can I let the webmaster know that I wanted to add a new page? Or to change something. So I just like, why don't I just put on a task list, give it access to the task list, have it check it a couple times a day. Right? Yep. And and kit and kit's the main bot, the chief of staff, is the one that kinda guided me through a lot of this too, right? It was it was pr is saying, you know, why don't we set up this, do that, or I tell it, create all these different bots. but it doesn't take a lot of work. But I think you really have to think about what area if I had unlimited resources to hire people. Like remote workers. That's probably the best way. If I if I can hire a bunch of remote workers to do work for me, what roles would they fill? And how s much can I simplify those roles to give them these different jobs? Right. Yep. Yep. I it's it's fascinating. You know, I I can relate, I you know, the the late latest agent that I developed was within twenty minutes, right? And it it had like four different sub sub agents that was governed by a main agent. that is the decision maker, the coordinator, things of that nature. So I I can relate completely. But it didn't take that long because when you have an idea in your head, it's hard to understand like w how do I translate that, right, to the to the world that is an agent. You just start talking to it, right? Just like you did probably for an hour, you're like, here's here's what I'm thinking. And it kind of fills in the gaps with assumptions. And again, you have to review it, but I think you bring up a good point. It's it's it's approachable. It it's not it's not out of reach. You just kinda have to understand what the parameters are and what the limitations are and what you don't want it to have access to, which is what you brought up earlier, so that you're not you're not giving away the keys. You're just basically ensuring that these agents are going to take action on your behalf in relation to how you would, but do it in a faster manner so that again, if for an individual it's going to be fantastic because you're not spending time l you know with a to do list that's just growing and there's no way to get out from underneath it. It's hey, Let's hand off 50% to 65% of what I should be doing to an agent and let me be more the orchestrator to a certain degree, right? So I I think that's I think that's a great use case. That's awesome. Yeah, no, I love it. It's making me I I I don't I we'll see how it goes after a few weeks of trialing this out. But so far, I mean it's been good. So but yeah, cool. What else you got? What do you got next? Well, it's funny. I I'm gonna tri segue into what kind of a follow up to what you just said there. And what's interesting is I'm noticing there is a story under underneath everything going on right now, which is everyone is talking about agents. And just like we said, most can't explain what that is. And I and I'm not saying nobody knows what agents are. I there is a lot of definitions that you can get get in there. But one of the things that I will say is the fact that an agent is a tool that does a task for you, such as like logging in, filling in a form, or sending requests. Instead of answering only of only answering a question, but the workdays AI at work, which they did a survey r recently and released on October twenty second of this year, found that only thirty-eight percent of workers say they can confidently explain what an AI agent is. Leaders were about two point five times more likely to be able to understand it, but then on a release on the twenty seventh from the Bankers Association, seventy seven seventy-seven percent of people are concerned about agents moving money. Nineteen percent say the agent had already asked for their credentials. So like It's getting smarter because the goals that they're setting within the design of said agents or the lack thereof, let's just put it that way, they're not understanding. It's not the agents that enable the I guess you could say the impactful nature of what actions it takes. It's us not paying attention to the details to give it direction on what it shouldn't be doing. So this this is the new wave, and it's funny. I noticed that the intelligence in itself. The workday data, 80%, 88% of the leaders had used agentic AI compared to 54% of the workers. So what we're seeing is the leaders are trying to get ahead of this, and sh they should, but the problem is I don't even think they have a good idea of what that agenc future looks like for their company right now, because they're still experimenting themselves. So I want to share some real data. And what I mean by real, I'm talking about humans that I interacted with over the last week. That was really fascinating. We had 19 people participating in what I called an agent design workshop. We weren't enabling the agent. We were just designing it in an impactful way. And we were following the principles of standard architecture of agents that Anthropic, OpenAI, Gemini all follow. So it's the standard practice of of AI. What was fascinating is I set the challenge up saying, okay, I want you to come up with a personal option and then an industry specific option. And I want you to exercise through this guardrails that I set up within a skill of how to effectively design an agent. So the first run was practice. They did their individual one. Everybody seemed to get it. Nobody actually had an agent though. Everybody had designed a workflow. And I thought that was fascinating because it's like everybody's still stuck in the days of automation. They're not thinking outside the box. So then I said, all right, now that we know kind of the parameters and we have a little scoring matrix that comes up to say, here's here's the three pillars it scored it on. This isn't an agent, this is a workflow. Here's some things to think about that you could use to make it an agent, if in fact you want it to be an agent. Here's the kicker. So we had everybody submit it. And I had I had agents doing all the design, scoring. I I was not the one doing it because I wanted to have an unbiased view of this. Sixteen of the 19 people that submitted the industry-specific agent had not decided who checks the agent's work. They didn't know. They just basically said, I don't know. Or they or they just let the let the skill design go without human in the loop, just like we talked about. Fourteen of the nineteen could not say what it would save or Or what it would provide. There was no value. it was just like, this does this. Well, what is that improving? They couldn't say. Thirteen of the nineteen tested only on made up examples. So they didn't have real examples for this because they didn't know where to find it, or they didn't understand how to leverage the, you know, the massive amount of data you can pull in through the LLM to be able to test it. And the last one is the biggest one. Twelve of the nineteen never ask the people who have the problem. They just basically assume this is something they want to solve. So what I'm getting at is It is in fact something that people struggle with is what is a work it's and we went over this with our group is what's an actual mechanical automation, what's an AI workflow, AI agent, and what's agentic AI? And there's those four pillars are important to understand because one of the things you're going to be challenged with is do you really need an agent to enable what you're trying to do? Because sometimes that's not the case. Simplicity is key. So I wanted to come back to the story you said, Eric, because this is really interesting. You were f almost forced to be able to do your job of developing these agents because if you didn't, you get way behind. So here's the here's the thing. It's the same thing here. When I developed the the agent I just did, it was forced because I had to get something in there because we didn't have that to be able to keep work moving forward. I think that's going to be the new wave of how people develop these, whatever they are, workflows, agents, agentic flows. It's they're gonna be forced by either removal of people, maybe it's the speed of work is is starting to tick up, and maybe even just simply, I don't have the capital to invest in FTE. I need somebody here to help me right now. And I think that's gonna be the movement of this next coming up calendar year, is people are gonna realize that being forced to design these things is a good thing. Because now we're thinking outside the box, and now the evolution of the market's gonna be, hey. Humans are going be able to interact with these things, but we're going to design them in a way that humans aren't focused on the administrative, they're focused on the decisions that need to be made throughout the process. Totally. So, you know, anyway. And that that makes me think of so you know, we we've been looking at hiring, as as everyone, as listeners know, looking at hiring a new role here at for Eligio. And it's interesting how the conversations I have with them, and one question I've been asking them, and these are just like We're doing these fifteen minute, I call it some of speed dating in a way, right? Because you get so many candidates and you do fifteen, fifteen, fifteen minutes. But I always I I always ask every single one of them, what do you think of AI and how are you using it? And I've had I think six interviews in the last two days. And just before we started recording this, is the first time that sixth person actually has some intelligence. and experience around what AI means. She really understood. She even talked about privacy. She talked about how the it's you can't just throw stuff into a free chat GPT account. As I was talking earlier, you you shouldn't be using free stuff. You should be using paid services. Otherwise, don't get me wrong, every everyone could get hacked. If you think you're protected in Google or Microsoft, y you are at least legally, but I mean nothing they could get hacked. But at least she understood the logic behind a free version and when you're paying for something, exchange of goods in in legal terminology. I thought that was very fascinating just to hear how most people couldn't articulate it. And e even she couldn't get it to the point where you and I are talking at this level, but it it's fascinating how many people are still workflow driven, for sure. Like it's yeah. I think how do you how do you get to the point where I I think and you say this all the time, outcome based, right? How do we get to outcomes? Right. What's the what's the outcome you want? And if I think of all those agents I was talking about, what's what's the outcome that I have for the webmaster? Well, to update my website. That's it. Yep. Yep. That's all. That's all you're doing. The SEO, what's their job? To identify issues with the website in room for improvement to drive more traffic. Yes. That's really that's really that's the job, right? So Simplicity. Simplicity of the job is gonna help you a elaborate and even tie them together to make an agentic ecosystem. That's the thing that everybody t seems to get lost on is agents are intended to have multiple roles. That that's not gonna now again, this could change next week. I'm not saying this won't, but right now i i if you simplify the job and simplify the outcome, you're only gonna get better results because then you hand it off into like you you have like a coordinator of some degree and you say, Hey Here's the outcome, what do we do next? Right. And it has that logic to be able to handle it. So yeah, I I just it was really enlightening and honestly, it was just telling of the need for what we talk about here at the podcast, but also the fact the work we do in the background of hey, we we're here to help you identify how to measure the c the output and the outcome with a f with impactful design instead of just throwing agents out there to go, Hey, this can manage your email. Well, what is it actually doing? Nobody knows. Totally. Exactly. Okay, cool. Alright, well let's let let's finish up on the last one that I have, which I guess we're gonna call Scam Alert. Scam Alert. Okay. Scam alert So it sounds like a sounds like a cartoon character, Scam Alert. Scam alert, scam alert, scam alert. so this so this is actually right straight from the Justice the US Justice Department's website. there's obviously news outlets too, but this is straight from the horse's mouth. The US Attorney's Office in Southern District of New York says three foreign nationals have been arrested and charged with a scheme to use they say super intelligence, but I'm gonna say artificial intelligence, to defraud companies of $29 million. So this is what happened. The the these these four nationals, what they did, they set a very sophisticated system where they had Phone calls originating from the CEO, so they had their voices and everything, calling lawyers and also mimicking lawyers, fooling the actual CEOs. those are extremely sophisticated, basically tricking people. The CEOs thought they're talking to the real lawyer, but the real lawyer was actually an AI with the voice. And the other the lawyers thought they're talking to the CEO, but it was actually the CEO with an AI voice, and they were able to push around $129 million. So this is a classic new example of fraud that people would have attempted in the past, but now they're doing it with super intelligence and mimicking and tricking people into moving funds around. So pretty crazy. And I'm glad they caught them. So that tells us the justice systems are working. But it's it's something to be aware of because this is a big number. And if you're a small business owner and it's like fifty grand. that that's I know it's not a hundred twenty nine million, but they're probably not gonna look at you the same. You're gonna be dealing to your bank. It's gonna be a painful experience because I've dealt with this stuff in the past. and you just won't get the attention that a big number would get like that or have the resources to go after that kind of thing. So it's very important that you know you start thinking they advise this, the Justice Department advises finding ways to put protections in c in in place. And it could be like a safe word that changes on a regular basis, right? Like Mike, if you call me up, I need to send money out to someone. Only I only do it if it's like I I don't know why this word came on. Purple Sunday. Maybe that's the phrase. It ha if you don't say purple Sunday, but next week it'll be a different new word and it just changes, right? it's do you know what I mean? So people need to say it gives a whole nother meeting to safe word, right? A whole nother meeting. my goodness. Wow. that's that's crazy though. Yeah, right. We went there. But like that's that's the crazy part though, right? I mean it's it's insane to the the amount of intelligence you can provide it to be able to do something like that, but then not think about the repercussions that could be related to when it takes it on its own. I mean, that's like going back to the basics of do you really want it to be an agent taking action on your behalf or do you want it to be more of a workflow? Like it's that's kind of like the balance you gotta meet and I you know I always think when you brought that up, the first thing I thought of is all those robocalls were so easy to identify that it wasn't a real human. Right and then it got better where humans would pre record their voice and still not great because they wouldn't respond to what you're saying. They just start talking, you're like, Okay, I already know who this is. But I have been on one of those calls, not a robo call, it was actually a r AI interview that basically they were asking questions and then they repeat back. Now I could tell it was AI because of the way it which it paused and then it would repeat back very strangely and you're like, Okay, whatever. Not gonna correct it. It it's accurate, but not, it's not human. but it's the same thing, right? Is like most people that don't interact with this kind of stuff, they're not gonna know. And they're gonna just be like, maybe this person's just talking a little slower, or maybe it's you know, maybe it's somebody that's very rehearsed. Like, who knows? But that that's the scary part is people that don't live in this world, they they they will don't know what's going on. No, they don't. And it's gonna be again, I think of I think of things like we've talked about this before, my mom, she's 82. yeah, and it makes me wonder how easily can she be tricked, right? So you know, I I do I do wonder about things like that, and you know, I I don't know what the answer is, but yeah. Okay, all right, looking at our time, because we're having great discussion today, but let's today's AI pop quiz. Why don't we Okay? I I didn't really prepare anything, but and as listeners are probably starting to figure out, I don't really prepare a lot of stuff. We don't prepare a lot of stuff, we just wing it. but it seems to work out, and hopefully, listeners, you enjoy that. So we're gonna do is I'm gonna share. Let me open up Claude. And so and you see my screen there, Mikey? yeah, we got it up. Okay, cool. Coffee and Claude. Coffee and Claude time. Yep, here we go. So we're gonna do a prompt. Please prepare a five question quiz based on Whoo, what do you want to do? Recent events in artificial intelligence in the last week? Yeah, yeah, why not? In the last week. Yeah, let's let's put a pr time time constraint on that because that could be a low difficulty. There you go. Make it multi make it multiple choice with a single response. Alright. Good? Love it. Alright, let's go. Let's go. Okay, here we go. Searching the web. See what's going on here. all right, November third to tenth, quick quiz. Remember we did this last when did we do this at the AI conference in September? I think we did pop quiz last. Yeah, we actually had content to play out. Yeah, we had content. This time it's all over the place. Yeah. Okay, first question. it's doing flashcard stuff and all that stuff. That's cool. Nice. what did Google announce it would do with Gemini for K twelve student accounts? Number one, turn it off by default. Number two, charge students a monthly fee. Number three, make it mandatory in all classrooms, or four, replace it with Chat GPT. I think I think this went like polar opposite on the the the s the spectrum of easy. I'm gonna be honest with you, 'cause the first one it seems like the most rational. Like turn it off, right? But I don't know. That's that's just me. I feel like I think I did read this article. I feel like I think they said that it's gonna be on It's mandatory. Really? I think they said that. I could be wrong. I feel cause I feel like it was like I think the I I know I was I admit I was headline reading on this one, but I'm pretty sure 'cause there's controversy 'cause all the the edit all the s the the schools are like, well we're have to take this out. So I gotta go with make it mandatory. Okay. But what do you think? I I didn't see this article. I'm just thinking about at the top of my of like what I think would be rational, which would be turn it off by default. But I now you read the article. I don't know. Well we're gonna do it. Nailed it. look at that. After criticism. Aha. So we're both right. After criticism from advocacy groups and senators, Google said on November 3rd that Gemini and AI in search will be off by default for K twelve students. That was only seven days ago. All right, so that that's why. Interesting. There you go. Alright, which company released a one trillion parameter open weight model large four on Hugging Face? I don't know this one. I don't know. I don't think it's Apple. I Klarna's like a f financing company. the only one out of this is Mistrol, which is actually an AI company. There's gotta be Mistrol. Apple kinda, but I don't think Apple's doing into the open weight models. Mistroll? No. No, I th Which one? What do you want to do? Which one do you do? I don't know. You pick this one. I picked the one. I'm go I'm gonna go I'm gonna go Mistral. Correct. Mistral released large forest open weights of one trillion parameters under a modified MIT license with a separate agreement for model as a service resellers, which I would assume that means like Hugging Face, for instance. Yep. Yep. Okay. All right. Next question OpenAI's AI agents found and built exploits for nine hundred and forty eight security flaws in which piece of software? And the options are Google Chrome's dark mode, Microsoft Word, the Linux kernel, or Minecraft. These are out there. I'm going with I'm going with Linux. It's gotta be Linux. I that's that was the what my copy Google Chrome's too specific. Yeah. Microsoft Word explode source. Yeah. Minecraft seems a little out there to be honest to me, but it's who knows? But yeah, I think Linux Linux would be the one. Right on. Open AI reported its agent pipeline found in patch nine hundred and forty Linux kernel privilege escalation bugs in a single run costing one point two million dollars. It followed with GPT six cyber, a model rated critical for cybersecurity capability. Cool. I mean, that's worth the money, right? For just what what could have happened, so I you know. That's that's the good news. Next question on its Q three earnings call, how many rides did Uber say were booked by AI agents during its busiest week? I don't I I remember seeing something about the report, but I don't I don't w Yeah, I didn't look at numbers. I saw the re Yeah, take a stab. Let's do one million. That's that seems relevant for like how many people use it. much sm much smaller than that. Boober said only four rides were booked via AI agents during Halloween week. Out of roughly three hundred and twenty million trips and orders, a good reality check on the agent hype. I don't really want I don't want an agent booking. I mean I love Grombot, don't get me wrong, but I'm not sure about having it book stuff like that. Yeah. Wow. That's crazy. Huh. All right. All right, last but not least, what did what new feature did OpenAI's images two point six update add to Chat GPT? GIF creation, full length movies, hand drawn sketches mailed to your home, or three D printing? I feel like I saw something about GIF creation. But I could be thinking about Gemini too, because I saw it on there as well. that seems I doubt they do full length movies. That just sounds like it's expensive. And this is about images, so I you're right. GIF. GIF? Yeah, I think so. Got it. Release November fourth. Images two point six added GIF creation in Chat GPT codex and the API. And with generation OpenAI says it's fifty percent faster. So we got what, four out of five today, bud? Yeah, that's not bad. That Q3 earning one, that surprises me. I was thinking it would be at least, a little higher with the amount that'd get booked on a percentage basis. So that's really fascinating. Totally, totally. Yeah. Cool. All right. Well, that kind of brings us to the end here. great great topics today, I think. Yeah, no no no. I I actually appreciate it the more the these stories are starting to show the human side more. on the AI, the impact it has. And and and again, it's not all negative. That's the cool thing, is we're starting to see the shift happening. And I think that's something we'll follow a lot closer over the next couple of weeks is with the IPO showing up with the company starting to realize the com computation costs are just out outweighing the actual principle of of operating on There's some cool innovations coming and I'll be interested to h see what happens next, especially next week. Agreed. Cool. All right, listeners. if you haven't subscribed or followed us, make sure you do. And of course we appreciate you listening and we will see you again next week. Thank you. Thanks for joining us.