Rob McFalls Live on Shally's Alley

Hey people, how's it going? Welcome, you are live on Shally's Alley, last episode of June 2025. It is Friday, it's 1 PM, and it's a good day. It's a good day to have a good day, right, Bill? It's a good day to have a good day.

Absolutely. Listen, every good day you wake up and you're on the top side of the earth, that's always good.

Every, every— for me, it's like every day I wake up, I'm like, whoa, awesome. Um, so folks, today I have a very special guest, and, um, this is actually somebody that I have just met. We met 15 minutes ago, believe it or not, but I am thoroughly impressed with his background. Um, the man with two first names, William Malcolm. Former head of— or director, I think, is director of talent acquisition for the DOD Air Force. Air Force veteran himself, um, see, magna cum laude from Trident University. Went to work at Travel and Leisure and Savita and was introduced to me by one of my colleagues, Paul Norman, who is a consultant with Riviera Advisors, um, like myself. And I was, uh, I was in a pinch because our guest today had to, uh, bail because of an important company meeting. And so it was kind of last minute, and I asked Paul, hey, would you recommend anybody that might be able to, um, you know, join us? And, and Paul said that Bill was, um, available today. So he's like, let me introduce you. And I looked at his background and everything, I was like, oh my God, This is perfect. So Bill, Bill or William?

I go by Bill.

Bill. Okay. So Bill has actually been building for the last 2 years at least, or been working at least for the last 3 years, have been working on it in the last 2 years, has been building a comprehensive AI strategy for a cancer center, which of course impresses the hell out of me because I am a cancer survivor. I had a battle with cancer that I dealt with for, um, a little bit more than— yeah, a year, 6 months. Now I'm in that like stage of like, you know, getting scans and everything. So man, to me, I need anybody that works in, in cancer is, is a hero. So you're already my hero for being in the, uh, in the military, and then on top of that, like, you know, like a double, double hero for me. Um, but Paul was telling me that you have been doing some really interesting things with in particular ChatGPT, but also generally speaking, all kinds of other AI initiatives. And I was really excited because I gotta tell you, man, I've been trying to find real practitioners, not vendors, but actual people who are doing the job that have implemented what I would consider sophisticated AI. And let me qualify that by saying not just use ChatGPT to rewrite my job description, right? I mean, let's just face it, that's really not AI. So Paul was telling me that you actually do legit AI. In fact, um, he said that you, uh, one of the areas you, you worked in is you slashed onboarding by more than half, uh, reduced onboarding time by, by more than half with AI-powered workflows. So this is robotic process automation and the kind of promise that AI gives us. So I really wanted to dig into that, if you don't mind. So welcome, say hi to everybody, and I've got, I've got a question to start out with right away.

So, all right, well, hey, thanks for, for having me on. I'm super excited. This is a passion of mine, almost, almost an unhealthy passion. I probably delve into AI about 4 to 6 hours a day, including during my work day. I listen to it in the background. I'm constantly looking at it. I get information fed to me by leveraging AI. I tell it what to do, what you want to look for, exactly what I want to look for.

Oh, that's really good. You're gonna have to tell us a little bit about that. But I'm really curious about this AI-powered workflow, man. Um, what, what is that? What, what's one piece of that transformation that most TA leaders overlook but absolutely shouldn't? Walking through that, like what you did, you know, onboarding using AI-powered workflows. Is there anything that people just overlook? Or, and if you can't think of something other people overlook, that's fine. What was it that, that, that like was the, the change agent, the trigger that was like, you know what, this is, this is going to have to be—

Yeah, definitely. It's— so when you say AI-powered workflow, let's, let's make that plural, workflows.

Um, got it, right, tied together.

I use— I absolutely, I use more than a dozen different AI engines on a daily basis, and that's just the popular ones, the, the ChatGPT, the Claude, the Gemini, the Grok, the, you know, all of those. And then there's other ones that complement that. It's a great time right now because it's an AI race. Everybody is racing each other. And because of that, there is a lot out there for everybody to see. So using all of the different LLMs or the large language models gives me the opportunity to go out and create different powered workflows. And what I mean by those workflows is help me find a way to make the problem that we have more efficient. Help me find— guide me in the right direction so that I can take questions that the company has given me, that I can take direction that the company has given me and actually put it on a path to help guide me. So when you talk about our onboarding that we did, where we did— I want to say it was about a 76% reduction, 75%, 76% reduction in the number of days Yeah, to onboard. We were getting people scheduled for their initial appointment at a company in 10 days. Said, here's your offer. And they were scheduling themselves for their initial onboarding appointment in 10 days. We took that down to about 2.76 days just by leveraging and asking the right questions. We were doing all of that stuff manually, making phone calls to people. And here's the thing, let's be honest. People don't answer the phone anymore unless it's mom or someone in your contacts. You better answer if it's your mom calling, right? Or, you know, someone in your contacts that you recognize. People just don't answer the phone. So we leveraged different scheduling tools to make sure that they were accessible to people. We were sending people emails saying, here's how your onboarding works. Here's all the things you need to do. In some cases, that included between 10 and 12 attachments. So all these are warning signs. One, yeah, one, emails go to junk and spam. Okay.

We know that that's happened for a while.

Yeah, absolutely. Two, people don't like attachments because of the nature of the world we live in and the risk. So we exchange that by using AI to build websites that now we send people to the websites We have all of the information that they need. So now we've exchanged all of that with one small miniature link that we send them and we say, go to this link. When you get there, you're going to ask to schedule your interview or schedule your appointment. And here's all the onboarding that you need to do. Not only that, we were telling them, yeah, we were telling them, fill out these forms manually. Well, we used AI to build online forms to prefill. Yeah, to prefill fillable online forms. How do we create those? Guess what? I went to ChatGPT, I went to Grok, I went to others and said, make me a professional on Adobe Pro. I'm not an Adobe Pro professional, but I learned how to do fillable forms.

So the form is Adobe. Got it. Okay.

So the form is Adobe Pro. They fill it online, they click. You know, submit, it goes directly into a mailbox here at the company. That was the life changer.

So what I'm hearing is that, at least for this, maybe there's other, but for this, you did not buy a technology that existed from a vendor that was saying AI, AI, AI, and it's the same vendor that 2 years ago was saying diversity, diversity, diversity, and 2 years before that was saying semantic, semantic, semantic. You actually used technology you already have. Uh, Adobe Forms, for example, maybe something that it has, but some people will have DocuSign or whatever. You use that, you created an app and a website using rapid coding techniques through these assistants, these intelligent assistants, and you essentially created your own applications, apps. Okay, so That's huge because this does partially explain the mystery of what I've been experiencing. And that is that, you know, I'm using a lot of AI, 60, 70% of my job, but it's augmentative technology that I'm using to help me rapidly attain the same results that I was doing before. So I'm not doing anything different or new. I'm just doing it faster because of AI, and I haven't paid one dime more than the subscriptions that I already have. So I've got Gemini because I'm a Google Work— Workplate, whatever it's called, the Google Professional Workspace. Yeah, I've got that. So it comes with Gemini. I am— I have Office 365 for life. And so that comes with Copilot. I pay for Claude and ChatGPT's $20 a month, um, pro, not the like whatever super, you know, yeah, yeah, yeah. But those two. So I've got four essentially that I pay for, but that's what I'm paying for. And I'm paying for that to do everything from home cooking recipes to planning family vacations and work. So I'm sort of, you know, I would have, I would have used them anyway. Right.

Right.

So that explains why I'm out there looking for people that have examples of having implemented technology, and I'm not finding very many because the people who actually are doing it sounds like are doing what you and I do, which is bootstrapping existing processes by accelerating them using augmentation. Am I on the money here?

You are absolutely right on the money with it.

That's awesome. That explains so much, man. Really appreciate it. So, you know, I want to talk about a couple other things that I just sort of pulled from, you know, our— from your content and stuff like that. A lot of people talk about AI-generated job descriptions and chatbots, right? Chatbots, I wouldn't even frankly classify as AI because chatbots have been around for a very, very long time. And what we've done with them is we've made them more powerful. Because we've given them access to a lot more data. It used to be a chatbot can only answer the questions that you predicted people were going to ask and you programmed it to a logic tree. Now, it can get a little bit smarter, but it also poses a bunch of other problems because it's too smart sometimes or has too much content. But people talk about the job description and chatbot. You actually created a screener. So, this is a chatbot that asks questions instead of a chatbot that answers questions.

Yeah, so it's— and, and you'll, shall we— you'll, you'll understand this right away. It's just a custom GPT. Yeah, that's it.

It's a custom GPT.

I told it— I tell it exactly what to ask. So if you know anything about custom GPTs, you gave it the screening questions. Absolutely. You give it the knowledge, you tell it how to think. So I fed it all of the knowledge I put in. What are the company's values? What are they trying to accomplish? What type of person are they looking for? Now, that's just the basics. Okay? That's the very basics. Now, you want to get more intricate with it? Feed specific GPTs with Hey, you know what? The hiring manager is Paul Norman. And here's what Paul is like. Here's Paul's LinkedIn. Here's his resume. Here's the type of person he likes to hire for.

And here's the roles he hires for.

The roles, right? He's looking for a self-starter, someone.

And here's the screening questions that he uses.

Here's the stuff he uses to prepare. So you load all that in there and that once you load in the resumes of the prospective candidates, you've already got the prompts that are in the pre-built prompts. So you only do it once, right? Then you put the prompts in and you say, okay, answer all of these. You let the GPT run, it runs through the entire thing. And then once you get through that, then you, you, you follow up prompts, follow up prompts. Okay, this is great. As a recruiter, give me the applicable questions I should ask. Okay, then when those are done and you've got your answers, okay, as a hiring manager, based on everything you know, what should the hiring manager ask? So what you've done—

you're helping the hiring manager with the follow-up questions.

Well, you're helping yourself as the recruiter because you're getting the questions you should be asking that are relevant based on the job.

Does it then give you a report on the answers? Like, gives you some summary? Okay.

It sure does. Yeah. And then what you're doing is you're providing a fantastic service to the hiring manager by saying, here's what— instead of just asking the same stuff you ask all the time, you know, hey, let's start with what's your, you know, what's your favorite color? And all those goofy questions that people ask, you're giving them a specific set of questions designed on what they like, what the company is looking for, and what the candidate actually is designed for.

So you can basically do what a recruiter normally would have done. Manually, which is, here's the recruiter questions I asked. So, you know, you put together your executive summary for the hiring manager. And you say, Mr. Paul, I interviewed this candidate. Here are the results of the interview. Here are my observations with regard to their qualification and how they match to the requirements. Here are the comments that they made and the questions that they asked and the questions that you asked me to ask, right? And all that information. You give it to the hiring manager. And in addition to that, what you are doing now, which you didn't do before, is you're also suggesting here are some areas for you as a hiring manager to dig into deeper because they either were missed or covered or we don't have enough information. And their recommendations— before, we used to recommend the hiring manager to ask certain questions, but they were based on an interview rubric or scorecard. Now you're doing that plus based on the actual results of the original interview.

Exactly. This gives them a roadmap. It gives them a roadmap.

And not only that, but does it reduce the amount of people they look at?

Absolutely. 100%. Especially when you can build the GPT to have a prompt that says, here's all the information on this person. Go out. Go out to the World Wide Web. Find everything you can about them. Go to their LinkedIn. Go to their Twitter, X. Go to all these social medias. Tell me what you can find out about that. You know, when you're establishing rapport, whether it's a hiring manager or recruiter up in the beginning and you're wanting to ask them questions, wouldn't it be nice to know the person you're talking to on the other end? You may have something in common with them. Hey, we both grew up in Buffalo, New York. We're part of the Bills Mafia. Hey, we're, you know, we're both baseball dads. This is some commonalities you may find. And let's be honest, you know, In business, it's about recruiting, or yeah, about relationships.

Yeah, it's about relationships, and, um, this helps accelerate that. But also, I'm, I'm wondering, is there— this is just because what I've run into, you do, you know, doing some, uh, uh, rubrics and things like that— is there a risk of false IDs? Or like sometimes— so you, you say go out and find anything you can about this person, um, does it Does it make mistakes, or, or how do you prevent it from like somehow finding Shelly in India who's, you know, not Shelly Stackel and, and somehow associating that with me? And then all of a sudden the, the hiring manager's like, this profile says that Shelly knows how to do video editing, but—

right, yeah, that, that's great, that's great, because, because it's only as good as how you prompt it.

So the prompts could be perfect. Doesn't it still run into like someone named Shelly and it gets confused? Or is it—

could it— could. And that's why in your prompts you put at the end of it, I need you to ask me all these other underlying questions to confirm and reconfirm that what you have found is accurate, right? If there's any— when you're telling the AI If there's any doubt in your mind on any of this, stop. Put that information in a secondary paragraph versus a primary. I basically, when I prompt it, I tell it I need the primaries along with validated sources. That being the key, the validated sources. When you look at, you think about valid sources in AI, that comes to mind.

Tell me where you found the information.

Where it comes. But now with all the others, they give the valid sources. So a simple And what's great about the AI now, if you do a simple mouseover, it brings up the little thumbnail of it, and you can see, and you can hit it, get it, go, move along.

I've got, I've got two recommendations that I wanted to make, but I'm gonna put a hold on that for a second here. And, um, so I ask— not ask, I guess I instruct. I don't know, sometimes I feel like I'm telling things to the, to the AI, but like it's a, it's, I'm asking but I'm really telling. Um, I, I basically say do not use any placeholders because sometimes they'll use placeholder text and they don't tell you it's there. Um, so I do not use, if you don't know, put, you know, missing or right, whatever. Like do not use a placeholder. And then the other thing is I specifically tell it, I don't know if this is the right way to do it or not, but I say absolutely under no circumstances, no hallucinations ever. Do not ever infer, interpret, or make an observation that is not verifiable by some source. If you don't know, if you don't know, just leave it out. Because I feel like sometimes it thinks it wants to answer the question, it will sort of fill in the blanks, which is what these models are supposed to do. So I'm kind of telling it to be against its nature. Does that work? Am I doing the right thing, or me telling you we're going to do it anyway?

No, you are absolutely on the right path. Now, what I would add with that also is people are now getting very hip to AI. Okay, so when you're doing that, also include no em dashes. Okay, that's that little elongated dash on there. Automatically.

Can you please, can you please tell me how to get rid of it?

Because just put in, just put in zero em dashes. You have, and it's still not right. Now I will tell you this, I have to have it in every prompt. I have it in my custom GPTs as an education piece, but in every prompt, that's what I did. It's automatically in every one of them.

I put it in the model and I put it in the training, but I don't put it in every prompt. I, I've reminded, uh, so ChatGPT, um, Pro and, and Claude Sonnet Pro. ChatGPT Pro, I have reminded it at least 8 times during the middle of a project, like, hey, you put some em dashes in there, like, do not ever. I've asked you before, right? Commit this to memory, remember this, no em dashes. And it still continues to do it. Claude Did it like 3 or 4 times and I got angry and I did something weird. I'm like, so Claude, I need to tell you something. I've asked you multiple times to not use the em dash. Why is it still being included and how can I guarantee that it doesn't get included in the future? Claude came back and said, oops, you're right. I'm not going to include it from now on and it finally stopped. But I had to treat it like a 2-year-old.

Right.

Don't stick your fork in the outlet. Don't— I told you not to stick your fork in the outlet. Okay, how many times do I have to— how am I going to tell you not to stick the fork in the outlet so you understand not to stick the fork in it? Finally, it did.

So, okay. Yeah. And I definitely want to get to your— to the underlying piece of your question. But, you know, as you're going through those, with this being such an exciting time with AI, they're all They're all racing each other. They know this. They know that people don't want the em dashes. It's people like you, me, and the millions of other people out there that keep telling it. I think just right now, example, I do notice with ChatGPT, they still continue a lot of that. That's why you have to be extremely specific with it. I think eventually they'll write it into what it is, you know, GPT 4.5 or 5.0 is going to be coming out later this year. I think they'll eventually get there. But getting back to your point of with the prompts, it's— I tell it, listen, don't just simply agree with me. I need you to be an intellectual sparring partner, right? So play devil's advocate and spar with me back and forth. I need you to— let's get to the bottom. We're working together as a team. You're not working for me. We're working together on this.

Interesting. Does it ever go too far?

No, no, no, no. It actually goes right to the right amount and it doesn't really go over the edge unless I go overboard. Yeah, I, you know, as much as just keep prompting it.

So, um, I read that you did some, uh, auto offer letters. That seems like it might be a little dangerous, but I'm sure you've got, you know, guardrails on it. Can you tell me a little bit about the auto offer letter side of things?

So this is just very simple. This was just uncovering that the vendor that we were with does e-signature offer letters. We were doing it where we would send it to people, they would print it out, they would sign it, they would scan it, they would email it back to us. And yeah, super, super caveman type stuff. We just understood, you know, hey, they have the opportunity to do e-signature offer letters. The better part about it now, same problem that we talked about earlier, you can send as many offers to somebody's email as you want. But guess what? When it comes from an ATS, the chances of it going to spam or junk are very great. But when you can take a link that you send somebody and you can send it in a text message and say, congratulations, here's that offer we were talking about 2 minutes ago, or even better yet, while you're on the phone call giving them the offer. Okay, you say, hey, pick up that computer in your pocket and I'm going to send you a text right now. Click that link on there and that's your offer directly.

So it's not an auto offer letter, it's automating the offer process with e-signatures and stuff like that?

Absolutely. Got it.

Yeah, got it. What about the Zoom SMS integration? What's that about?

Yeah, so, so we— the There's a lot of scheduling organizations out there, you know, automated schedule, Calendly, GoHire is another big one.

So Zoom, I have seen that fail miserably in so many ways. I wanted to find a way to get automated interview scheduling working, but it has been problematic.

Yeah. So, so this one, the reason why it, it benefited and worked for us was we have an existing account with Zoom. The onboarding appointments that we do are done via Zoom. So it just made sense that we call Zoom, turn on the suite of Zoom Scheduler along with Zoom SMS. And now when we drive them to the scheduling link for the onboarding and we say click on this, it comes up, it built on a calendar across 4 people. It's a— it's called a one-to-many, meaning it's one person scheduling, scheduling people's calendars availability. They pick the available daytime. It's linked directly to their Microsoft 365 Outlook calendar. And not only when it's scheduled does it go to the candidate's email, it goes with the link, the actual Zoom link, right in there. You don't have to do anything secondary to it. That's really the driver behind us going from this 10 days down to the 2.76 days is because—

So for onboarding, I can see that being a little bit less complicated. Is it possible to do this with hiring managers? Because I ran into roadblocks and, you know, I don't know if you have a solution for this. If you do, that'd be awesome. I'd be like, this is worth— I should be paying you right now. But my scheduling robots, basically, they caused some problems. I'll give you 2 or 3 of them. There were more, but these were the ones that came up the most often. Number 1, the hiring managers had a lot of placeholders, have placeholders on their calendar, for example, blocks of time. So, things like heads down coding, no meeting Wednesdays, you know, casual Fridays, stuff like that. And sometimes were hard do not schedule anything like for 2 hours head down Mondays, you know, definitely not. But some of them were, you know, like, it's no meeting Wednesday, but it's no internal meeting. That doesn't mean no interviews. So they block internal meetings, but they still want to do interviews. And the calendaring system doesn't really understand that. It doesn't know that this is something you can overwrite and this is not. So that's one. Another one was hiring managers that didn't really keep their calendars really current. And would like get scheduled and be like, oh, I'm not here that day. So their day-to-day, week-by-week, that, you know, this week's calendar was okay, but a week out, if they had vacation planned, they didn't put it on their calendar. You know, booking things on their leaves. It wasn't as much their calendar is up to date today and tomorrow. It's the days off and things like that. And sometimes some holidays that are not national holidays and things like that. And then the third one was changes. I would have a hiring manager, and this is a— let me start by saying this was a cultural problem that we had to address, and it wasn't something TA was going to address. But we did have hiring managers who thought that they were better than that and wanted to reschedule the interview 10 minutes before the interview. So those are three scenarios where Automated interviewing, just kind of like I had hiring managers yelling and saying, stop it, we don't want to do this. It's, you know, just send them to me and I'll book it because they just didn't want to deal with all these changes. Can those be solved? And I understand the third one is literally a cultural problem, but can those be solved with AI or not?

So, you know, you could, in a piece of it, I got a couple different opinions on this. The first one being Can you use AI for it? I think about using Copilot, which matches with your Outlook 365. You can go into Copilot and like, listen, you can type in there, here's my upcoming vacations, or I'm going to be out of office at a retreat or whatever. You can tell Copilot. Yeah, you can hire a manager to tell Copilot.

Okay, so it's an educational thing.

Correct. And it will remember that. Okay, okay. Um, if you're that lazy when it comes to technology, but my initial thoughts lead down the road of there may not be that technology savvy, because what it's going to come down to is that they're just going to have to be responsible to manage it.

Yeah, which is a cultural thing.

I mean, which is— it is.

What about this? What about, um, what was— I was thinking about a interrogative robot. So what it would do is it would, it would essentially create, uh, the candidate's availability schedule. So it would ask the candidate, hey candidate, um, what are your hours availability over the next, I don't know, whatever, 5 days? And then it would select from the calendar, here's when I'm available. And then it would send that to hiring manager number 1 and say, are you available at any of these times? And then the hiring manager would pick one of those times. So think Calendly, but it sends the Calendly to the candidate. The candidate picks multiple openings, then it sends it to the hiring manager with what's remaining, and it says pick from these, and then it sends it to the next hiring manager. And then it has the ability of reply to me if you need to reschedule. So that way the recruiter isn't the one that gets the message saying, hey, can you reschedule this? It'll actually be doing it through email. Is that out there somewhere? It's not—

it is. It is. What you're describing right there is in Microsoft Outlook. And, and, you know, quite honestly, is Microsoft Outlook is a fantastic tool, and you can do that with an external candidate, not a— you sure can. It's just so— there's so much in it. It's really just getting the information out to people. And I'll tell you This is a little-known thing. A lot of things that drive my curiosity is Instagram Reels, TikTok, you know, LinkedIn videos. That really— those short snippets that give me just a— I want to know a little bit more, and then that drives me to a YouTube channel or to a podcast or to a webinar or something like that. So You know, it's, you know, how, how did I mention that one about Outlook just right now? I saw a TikTok two nights ago about doing exactly what you were talking about.

That makes sense. Well, that's what I was kind of thinking is it would probably be possible, but I just wasn't sure if it was because, you know, Calendly doesn't really do it. And I do like those online calendar tools for me because I can send a link and someone can schedule time on my calendar, but I have my calendar up to date and I still make mistakes. I mean, I booked somebody for 4th of July. And I, you know, hello, right?

Exactly.

I should have known better, but I didn't. Yeah, there you have it. Um, all right, before I ask you the next question, I told you I was gonna— you've given me like thousands of dollars worth of valuable consulting already. Let me, let me return the favor. I don't know if this is worth thousands of dollars, but it's probably worth at least 5 cents. So my new browser is this. This is my new browser. Notice anything weird about it?

Uh, no, just very basic. It's ChatGPT, right? Nope, nope, it's a browser.

It's a browser.

Oh, hold on, is this the one out of— is this the new Gemini one? Nope, it's, it's pretty small.

It's from the browser company, uh, and it's called Dia, like day in Spanish. Yeah, but let me just show you. Let's say that we wanted to do, um, interview scheduling for recruiting. So it comes back and say, all right, automated interview scheduling. All right, Then I can say, what are the best options for automated one-to-many interview scheduling in recruitment? What this does is this is a fully integrated AI that knows everything about me. So it knows all of my hard drives, my cloud storage drives. It knows my technology stack. It knows my email, everything. And it's completely like, these guys are super paranoid about privacy and security. And I double, triple checked. I work for the government. So I double and triple checked this. So it only uses information that I authorize it to use, but it knows me really well. And that everything that you do with it is a chat. So for example, let's say that I wanted to open that one, and I don't know anything about that one, and I do know Paradox, and I do know greenhouse scheduling. So let's see. Okay, are there any options for Email confirmation, one-to-many. Let's see what it says. Those are the same. Let's use Paradox because that's the chatbot we were talking about before.

Yeah.

Now here's the kicker. So all of these, what it's doing is it's not browsing, it's telling me about the features right now. I can come back here and I can say, uh, create a comparison table between Calendly Good Times and Paradox.

Now, now, as you're doing this, do you have licenses with those?

No, but I can. Like, I don't, I don't right now. It's just giving me the information. So let's say that I wanted to open Calendly, right? Yeah, Calendly. So I'm going to go to Calendly here. No, not calendar. Calendly.

And the reason why I asked about the licenses— yeah, yeah, the reason I asked about those is what— maybe just a difference of how things would be done. I would say, here's all the things I have access to, can you—

oh yeah, I have that myself.

Yeah, yeah, can you help me with this myself? Like the—

I'm just trying to, uh, show the, the browser capabilities right now. So it was Goodtime, uh, Calendly, what was the other one? Oh, Paradox.

Paradox.

It's paradox.ai, I think. There we go.

Okay.

Right. So let me rephrase that here and say between Calendly, Paradox, and GoodTime. There you go. So now in this chat, I'm mentioning the other tabs, right? It actually uses all of the tabs. It can reference the tabs that you have open and it does the work for me. Basically, it's changing the perspective from the browser is something that you use as a window to look at the internet to one of the browser is a prompt. And everything you do is prompt-based. Yeah, you see how that works?

Yep, yep.

That's, that's kind of like the difference. This is the old way. Yeah, and that doesn't work anymore because I have to go through all of that stuff, right? This is the new way, and that's what I'm doing now. And then you can also do these things called, uh, skills. Let's see if I can actually build one from here.

When you say it's doing, is this running like an operator style thing like with ChatGPT Plus?

It's its own. Yeah, it's theirs. It's proprietary because it's secure.

But yeah. But is it running? Is it like operator where it's doing the work for you on each of those tabs?

Yes, but it's not connected to ChatGPT unless you wanted to.

Right, right.

You would have to tell it to do that, otherwise it's going to use the, the browser company's one, right? Um, and this is the other thing, is the skills. You can program skills, so you basically can create a skill that does whatever you want. And so this is what you were talking about here, right? So I could make a prompt that, that does the work when I use that skill. So this could be, for example, scheduling. I haven't done it yet, but I just could do scheduled, right? So, um, book a meeting between people in this email using Calendly. Uh, you know, avoid the M-, uh, don't book whatever. So the, the prompt then becomes permanent as a skill. You never have to type the prompt again. I know that seems basic because like, oh, you could save your prompts in a prompt database, or you can tell, you can tell ChatGPT using projects. That's the, um, that's the main thing, you know, during projects and stuff like that. But this is different in that it's kind of like built into the software, and all you have to do next time when you do this is just type in the prompt for schedule. I would have to be in my Gmail, and then it would know because I'm in Gmail, look at the people in the email, and then send them all calendar requests using Calendly. Does that make sense?

Absolutely. And you know what? It's like they'll eventually, just like ChatGPT with connectors, they'll get connected. All of these AI companies will get connected just like—

Yeah, yeah. This is kind of the bridge right now.

But it'll connect your Gmail.

To me, the biggest change that I think you would enjoy is the fact that the browser is no longer a browser with favorites and bookmarks, and it's basically a prompt and everything starts with do this or whatever. And it will— it does have your bookmarks, it remembers all of them. But you don't have to save bookmarks, you don't have to, you know, because it tracks everything in the backend and you can just automatically do it. So I thought you would enjoy that.

Um, well, absolutely, it remembers, it remembers, you know, hey, you, you searched for this, you know, this while back, you know, um, I'm gonna bring that up again. Or yeah, yep.

Uh, one of the other things that I found on, um, on on the, you know, the, the projects that you were listing is you have a requisition authorization request. Now, I built something for this in Workday a long time ago, uh, 3, 4 years ago, and it definitely wasn't using AI, and it was a big deal. But it sounds like the one that you did is, um, pretty much kind of a, a workflow or multiple workflows for position authorization using AI, and it It said on your, on your project listing you saved $12 million.

We sure did, par levels.

So how did you convince people to trust, uh, the, the process of requisition authorization to a robot?

So it's, it's, you know, a lot of it was just based on letting them know, number one, focus on your vacancy, okay? We're not going to scrutinize you having to fill replacement positions. Those are well within budget. Okay. What we will scrutinize is growth. You know, do we have the ability— you know, the company knows if they're going to be able to grow based on obviously revenue. So with that being said, it's really just programming and saying, okay, we know what our growth is anticipated to be. We want to make sure that we are managing that, but by letting them know upfront, your requisitions shouldn't go over this amount. And that even feeds down the line. We get super specific. Hey, you should hire this many per month, even this many per week. So keeping them on that straight line, you know, and these are for a lot of the ones that don't have like Workday, like position management, position management and Workday Workday, super simple. You have a position that goes vacant, guess what? It's going to ask you, would you like to open up the requisition? Very—

the problem was that they were just opening them willy-nilly and they were giving them different names and not using standards. So we ended up having to create a position management, um, application that I used back then. What I ended up doing is I used, uh, robotic process automation using Microsoft Forms and the Microsoft RPA tool, Power Automate.

Power Automate. Yep.

And then, you know, connect. So that was like early on AI in the sense because it was more like automation. But yeah, that was challenging because everybody wanted to copy and paste or use the same thing from last year and things have changed. You know, like we have different codes and all that.

And you know what, a lot of people, a lot of your viewers might be scared. I don't know what to do with Power Automate. Listen, I wasn't far behind, but when you can use Power Automate today with Copilot sitting right next to it and saying, okay, here's what you do next, here's what you do next— as long as you know how to use the AI, anybody can be a professional on this stuff. You don't— no longer— you no longer need a Microsoft 365 expert to do all of this.

Yeah, and you no longer need the RPA stuff because that was a little clunky. That was absolutely clunky.

It'll teach you how to code. I built my son— I was manually putting stuff into ChatGPT for him to learn vocabulary words and then printing it out on my printer. To me, that was primitive. So what did I do? I put the vocabulary words into ChatGPT and said, give me this format, which is the same as he would take the testing. Give me a practice test. And then I said, okay, here's what I need to do. I need to put it on a website. It created me a GitHub account. I created a website. I sent him the link via text. He was coming out of baseball practice. On your way home, go to this link that I texted him. And it was— he filled in the answers to the questions and it sent the responses back to me. All of that in like 12 minutes.

And the link was a a website created by the robot?

The link was a link, was a, was a link to the website that, that Git— it— ChatGPT helped me create that via GitHub.

Nice.

I've never used those before in my life, and in like 12 minutes guided me right how to do it.

And the website, was it hosted somewhere else, or—

it was, it was hosted on GitHub, so it was like a little—

oh, it was a GitHub website. Okay, got it, got it, got it.

Okay. A simple website, just simple.

Yeah. So Paul is asking me a question for you. He wants to know, how do you get the leadership team, the executives on board with AI because they think it's scary?

You know, I think a lot of it is through demonstration. We were talking earlier about— I mentioned very, very early on in the podcast about news. And what news do I use? This is one of the key things I tell my AIs, and I use a lot of tasks, which, you know, tasks seems like a, you know, to you and I, you know, we may think or to your viewers might think a little bit differently, but AI tasks you can build to where it drives the information to you, but very, very specific. And I'll tell you what this looks like. Years ago, I would go to Twitter, now it's X, and I would put in, here's the stuff I'm interested in. It learns through algorithms what you're interested in, just like TikTok does today. When you can go into the AI tasks and say, I need news every morning that is relevant to a large cancer center. I— here's what I need you to know about that cancer center. Here's all the things that we're currently working on. Here are the goals we look to accomplish our next fiscal year. Here's the goals we look to accomplish in human resources. I need you to go out and find news that correlates to that across the globe first, second, across North America, third, in the state of Florida. I need you to correlate all of those things. I then need you to give me summarized talking points that I can provide to the chief human resources officer and his peer cabinet members, talking points for them to speak of. And then lastly, give information for his direct reports leadership team from there. And I have it produce that to me. I weed through. Slim it down and provide that to the chief human resources officer.

Is that on a web page that you subscribe to, or like, is it just the chat— the robot gives you this every morning?

Whatever the, the AI is, whether it's Grok or ChatGPT, I have multiple ones feed in because I like my sources coming in from everywhere. I just don't, you know, and they all feed into like your inbox, basically. They all feed into my inbox. Okay, so what I normally do is They email them to me, and then normally what I do is I take them and I put it in my most preferred AI engine, which is ChatGPT, and I say, okay, now, or whatever, take all of these, remove, remove any duplicities, and then from there, give me conversation points that I can provide to the CHRO so he can spur the conversation from there with his leadership team. It's not rocket science. It's really just knowing how to use the AI.

Prompting the search, prompting the correlation, prompting the summarization, and then you finally are the arbiter of quality.

Right. And you know what the difference is? I mean, it's a difference in, I guess, call it generations, whatever. My brother-in-law, he's a He's a Wall Street Journal guy. You know, he'd be like, I got to see the Wall Street Journal. And he knows what's published in the Wall Street Journal. Okay. This is my thing. I'm like, give me world from across, give me news from across the globe. But not only that, but it's relevant, relevant to what I'm doing here today, right now. That's where the rubber meets the road. These are the goals we're looking to accomplish at this facility at this day and time. Give me the news that's relevant on what I need to be ready for. And that's really where it's given me a lot of the ideas and a lot of push with different AIs, because I ask it, give me the different AI things that are out there that we need to be prepared for.

So build our own personal news engine, basically. Absolutely. Your personal news engine for the use of the company, for the company to use.

Yeah, correct.

Okay. So to wrap things up, uh, last, uh, last question, um, that I had for you, and then we can get to some of the— there's, there's a couple things I do when I close, um, and this one comes from Paul as well. Why, why do you think TA teams aren't using VO to make Bigfoot recruiting videos? So He showed me the gorilla one, by the way.

Yeah, yeah, yeah. Veo's new. I've used it myself. It's not perfect. It does come down to prompting.

But you could build a fun video. Is that not the right, like—

No, no. I would absolutely do it.

Why not?

Absolutely do it. Now I think companies would be a little apprehensive.

Oh, right. Okay.

You know, oh, is that our branding? The branding, yeah. Here's the thing. There's a, there's a guy. And you know what, on LinkedIn, he did, he did an AI video for— and it was aired during the NBA Finals. PJ Asciutturo. Okay. PJ Asciutturo. And he's a guy, he's got a little bit of a marketing background, but he's not a— he's not by any means a production designer or anything like that. He did. He was, he was, um, asked, and for like $2,000, he did an all-AI video with VO3, and it aired during the NBA Finals.

Wow.

Yeah. So if you're ever interested in seeing that, it's— but that's where we are right now. I mean, how— think about how long VO3's been out and when the NBA Finals was. Um, the timing was, it, it It's crazy. It really, really is.

Yeah, that is.

Yeah.

So first of all, thanks a lot for taking time today, especially last minute here. Um, it's— this has been extremely educational. I do have two questions that I ask every guest, and I know you're kind of new to this. Um, you didn't even know who I was before, so, so this might be a little bit of a surprise. I'll give you some, um, I'll give you a little bit of time to prep, but the two questions are this: First of all, who should be our next guest? Just like I found you through Paul, who would you, who would you like to see on here, or who would you like to be interviewed, uh, or someone that you just think, you know, you highly regarded, respected, and, and should, and had something interesting to say for, for our community?

You know, if you could— I'll reach out. I, I just mentioned PJ Satoru.

Um, okay, that's true.

Listen, if you could, if you could get him That would be on the heels of the conversation we just had. I think that's, that's going to— that's the next big thing. Here's the thing. When you think of video and stuff like that, it's everywhere. I mean, that's the thing. People are into the small little snippets of stuff in the palm of my hand. Here it is. If you could get PJ, that would be fantastic as a backup. To that, hmm, I'd have to think about that. Maybe let's go to your next question then.

Okay, all right. The next question is sort of, uh, a philosophical one. A lot of folks read the transcription or view or watch this over the weekend because they can't come on Fridays at 1, and so I want to give them something to, to think deeply about for themselves. Um, so I ask all my guests to give us a question that we should ask ourselves introspectively, you know, after we listen to the show? Something that we ask— no, give us the answer, just give us the, the question we should prompt ourselves with.

This is an easy one because this is not only a question, this is a question that I started asking myself, um, when I became an Air Force recruiter. And for those people that may or may not know me, I was the most decorated recruiter that the Air Force has had. Okay, that's something I didn't— you can always just say thank you. I will toot my own horn there, but you should.

Goddamn.

Yeah, I always ask myself, if you're not trying to be number one, what number are you trying to be? You know, coming into work, at home as a dad, at home as a husband. If I'm not trying to be the best husband possible, the best dad possible, what am I trying to be? Am I just trying to—

am I just trying to be on the list?

Or am I—

I'm not going to answer that right now because I have to think about it. That's a really good one. That's a really, really good one. Are you— are you shooting for number one, or are you just shooting for Top 1,000, or are you— do you not even want to be on the list? Yeah, because you got to plan for it if you're going to do it. That's awesome. Absolutely, dude. This was absolutely fantastic. I feel like you and I are like brothers from another mother. Um, I, I definitely have the passion that you have for AI, so we did— we should keep in touch for sure. Absolutely. I'd love to have you back on the show. If there's anything I can do to help you with your AI initiatives, don't hesitate to reach out. And, uh, man, yeah. So folks, for everybody, um, that's out there, if you're interested in coming back, uh, Friday, it's the 4th of July, but we do have a guest. Our guest coming up is, um, the, um, uh, on the 4th of July is Adam Kovacs. He is from the Talent Foundation, so it should be an interesting one. So we'll hopefully see you back here tomorrow— not tomorrow, next Friday for Shallies Alley. Until then, be kind to yourselves. Bye, everybody.