Latest / Recruiting Evolution / From Cold Call to Closed: AI for Recruiting Firm Business Development

Transcript

From Cold Call to Closed: AI for Recruiting Firm Business Development · Recruiting Evolution

• Keely Flood [00:00:00]: I feel like more often than not, this AI spam outreach stuff is gonna hurt your brand more than it's gonna help. It's gonna burn bridges instead of build them. I'm not technical by any means, but when Claude Cowork came out and I saw the simplicity in using it, I thought, okay, this is something I need to pay attention to and really learn how I can get better at it. Kortney Harmon [00:00:21]: Welcome to FDE+, where the conversation shaping the future of talent continues beyond our live event series. This special series brings our free virtual event sessions directly to you. So they're featuring thought leaders, top trainers, experts that are really tackling issues that matter most across the recruiting, staffing, and talent acquisition industries. These sessions originally happened live with real conversation, practical insight, and open Q&A. And now we're bringing the best of them to the podcast. Listen in. We had a wonderful first session. Keely, no pressure. Kortney Harmon [00:01:05]: You are session number 2 in day 1. And obviously, as we keep the momentum going, our next session is From Cold Call to Closed: AI-Powered BD for Recruiting Firms. I'm super excited, especially if you say you're gonna share Claude live. That's a little fun. Keely Flood [00:01:21]: Yeah. Kortney Harmon [00:01:22]: So if you have ever started at cold outreach, like Mark was talking about, and figured out where to start, where you need to go. This one's for you. Keely is the founder of The Honest Salesperson, and he really brings a different approach, no BS, to sales and business development. And I'm really excited for him to share how AI can actually help you move prospects through your pipeline faster and more authentically. So Keely, with that being said, I'm gonna let you go. Let me answer one question though. This question is, is this going to be recorded? Yes, this is being recorded. You can reference this in the same link that you registered for. Kortney Harmon [00:02:00]: And this will be dropped on the podcast of The Full Desk Experience as well. So it is multipurpose, repurpose, share with your teams, all of the things. So, all right, Keely, I'm going to add your screen. I'm going to jump off of here, holler my name, and I'll come back on. It's like a bad genie that just comes back. Keely Flood [00:02:16]: Can I just start yelling help, help, help, and you appear? Okay. Kortney Harmon [00:02:19]: 3 times. Keely Flood [00:02:20]: Got it. Perfect. Well, then I will. Yeah, I'll get rolling. I know. So I see some people. Yeah, they came back, large cups of coffee. All that stuff. Keely Flood [00:02:28]: And so today, obviously, I want to talk about AI, Claude, hopefully with some time, do some live demos. I think that's been really impactful for what I've been doing this talk is just to see it in action. And before I even like really get into the full slides, I always put like a preface around this with AI in general. I'm a big believer in it and to an extent. And what I want to really cover today is ways that I've been able to essentially buy back my time to do other things within the recruitment space. Because my, my core philosophy and kind of what I believe in is in the recruitment industry, the 2 highest value activities, like literally the money-making activities, are talking to candidates and talking to clients and prospects. Everything else is secondary as far as I'm concerned. And so if I can find ways to get back 20 minutes of my day, 50 minutes of my day, 3 hours of my day to focus on those 2 of the highest revenue generating activities, I feel like it's a worthwhile endeavor to learn about it, to experiment with it, and ultimately to use it. Keely Flood [00:03:29]: And that's what got me to using Claude Cowork because, and I'm not gonna go through like a ton about, I guess I'll talk a little about me and I'll give a brief history about Claude and then we're gonna get into it. So a little bit about me. I run The Honest Salesperson Consulting, a consulting and a rec-to-rec recruiting firm. Over the last 3 years, I've worked with over 50 different recruitment firm owners on a one-to-one capacity. I also have run group programs as well. So the number of people I've helped is in the hundreds at this point. And my sole focus is on the sales and BD side of running a recruitment firm. I myself started in the staffing world back in 2015. Keely Flood [00:04:06]: I was a split desk 180. I was doing the fulfillment side. I always wanted to get into sales. And the moment I was able to move over to the sales side, I was kind of hooked and I said, hey, I love this part of the job. Fulfillment, don't get me wrong, it's obviously a vitally important part of the job, but I just love sales. And I also firmly believe that sales is one of, if not the hardest part within this industry. It's getting new clients, it's growing revenue, it's growing your book of business. And so what I do and what I have been doing is helping recruiting firms and their owners and everyone else get better at sales. Keely Flood [00:04:37]: So we're going to talk a little bit about why I think you're leaving hours on the table every week if you're not using AI. And where I'm going to give you the brief history of my kind of journey with Claude is back in January, I believe January 12th of this year, something came out called Claude Cowork. And if you've been following the AI space like I have for— probably all of us have, right? Because if you ever go on LinkedIn every day, there's someone saying AI is going to replace the recruiting industry, yada, yada, yada. They've been saying that for years at this point. And up until really January, I kept seeing that stuff. I saw the outputs and I thought, it's not really that great. Like, I don't see a ton of value in it. I feel like more often than not, this AI spam outreach stuff is gonna hurt your brand more than it's gonna help. Keely Flood [00:05:19]: It's gonna burn bridges instead of build them. But then Cowork came out, and I'm not a technical person. My background before I even got into recruiting, I was a college football coach. And if you were on just a moment ago, we were talking about— I'm back into high school football coach. Like, I'm not a software engineer, I'm not technical by any means. But when Claude Cowork came out and I saw The simplicity in using it, I thought, okay, this is something I need to pay attention to and really learn how I can get better at it. Because the ability for AI, and you see all these terms about agents, Claude calls them skills. It's basically AI's doing things for you on your behalf, whether that's updating your CRM, which is something I hope to show today, whether that's helping you create presentations, which, to be honest, the slides you're seeing today, I helped— I had Claude help me create this stuff where I didn't have to go in and do all of this. Keely Flood [00:06:08]: I gave it some ideas, I gave it the copy. It actually went and helped me create the slides you're going to see today and saved me a ton of time. And so I say that part because there are gonna be some parts of the slide where it's gonna be a little bit off formatting and things of that nature. I want you to see the full gamut of using AI. There's a ton of positives, but there's also gonna come with some negatives. It's not always gonna be perfect. It's gonna look a little bit weird, but I think it can buy back a lot of your time to focus on talking with candidates and clients. The other thing too, before I kind of get going, I'm going to monitor the chat side too. Keely Flood [00:06:38]: So if you're asking questions during it, or I shouldn't say if you are, please be encouraged to ask me questions as I'm talking. I've done quite a few of these now over the last 2 months in terms of like workshops. I did one with a group where I probably had like 75 questions in the course of an hour on the topic of AI. And I spent my time going through that. I'm obviously going to go through slides, but I want to make sure this is almost like an open dialogue as I go through this stuff as well. But we'll move on to this. It's not that I see ChatGPT like more as not disparaging OpenAI. I think their model is great. Keely Flood [00:07:09]: I'm more of just the way people were using AI until like Cowork came out and these agents came out was it was almost like a search bar. They'd go in, they'd say, hey, I need this, or I need that. And then AI went out and found stuff. It was pretty cool. And it would send some stuff back. But the issues you run into with using that like chat function It's, it's a silo. It's, it's a standalone chat. And every session you have to go in and you'd have to like reprompt that AI model to understand what you're trying to accomplish. Keely Flood [00:07:37]: It doesn't obviously know who your clients are. It can't do things within your CRM. It forgets everything from one session to another. So you're basically starting from zero over and over again. The way you can use Claude Cowork is you can turn this thing into your second brain. It can be taught to remember things such as who are your clients, how do you do submittals. It's connected to your CRM. It can go in and pull out data and say, hey, here's the people you have not followed up with in the last 30 days. Keely Flood [00:08:06]: It can do— and it says HubSpot on there automatically. That's the thing I use, so I want to make sure that I clarify that. Obviously with Crelate, I believe— and Kortney, you can save here— but I'm pretty sure you guys have now got the MCP side of stuff taken care of. So like, there's some really cool stuff you can do. Within Crelate, I just want to talk about my own world. I'm not trying to pitch HubSpot by any means. That's just what I've used because when I show the demo, you'll see it kind of live in action. But I know with Crelate, you guys are doing all this stuff too. Keely Flood [00:08:29]: So it's pretty cool to have Claude integrate within it, correct? Like, I don't want to speak out of turn. Kortney Harmon [00:08:33]: You're absolutely correct. Keely Flood [00:08:35]: That's my genie moment. Kortney Harmon [00:08:36]: We have— no. Yeah, that's your genie moment. Yes, we actually have all of that built in. I'm actually going to show you after this session a free trial. You can actually dig in, play with our AI. See it live in action, get your hands on it. So I'll put that in the link while we're chatting. Jeannie has disappeared. Kortney Harmon [00:08:52]: You are, you are not incorrect in your answer. Keely Flood [00:08:54]: Awesome. We did not plan that, her and I. It was literally organic. So like, I'm glad she was still around, not grabbing a cup of coffee when I need her help right then and there. So these are some of the like tools and skills that I have created. You're gonna see some things up top there such as Google Drive, HubSpot, Ocean, Notion. Those are my own personal tools. I give this slide as an example of how you, when using Claude, can set up your own— they call them like connectors— your own way to take your tools and integrate it into Claude. Keely Flood [00:09:24]: So that way, when I go in and ask it to do certain tasks, it will then pull up the necessary tools to actually execute on it. So for example, ocean.io is something I use for data when doing prospecting. And so I could go into Claude and say, hey, I want to go after these accounts. It can then connect with Ocean, find the accounts, find the leads, pull them out and give them to me, and then even automatically upload them into your CRM from there. The next bar underneath, those are some of the skills I've created. And again, Claude calls them skills. Most of the world calls them agents. How my simple coaching brain thinks of it is it's an AI. Keely Flood [00:10:02]: It's part of AI that actually does something. It creates something. There's an output to it. And so the prospecting BD, I'm gonna show here in a minute how you can use Claude to do some prospecting for you. If you do like coaching like I do, or better for most of you on this call would be post-meeting recaps or post-candidate call recaps. If you do have an AI note-taker, which most people do nowadays, you can have it set up where Claude will literally integrate within to it, pull the transcript, and create for you a recap email for you to send out on your behalf. When it's properly trained, there's last part there—the bottom part where it says auto capture—that's where the Claude MD file is basically like the memory file where everything can get stored into. So Claude knows where to go and remember things from session to session. Keely Flood [00:10:47]: I even have it set up where every week it auto scans all the stuff I've talked to it over the last week to then go and pull out information, capture it, and make sure I'm not having to reteach it from week to week from there. Again, my personal opinion where I think AI can be of help in every part of the cold to close stage. So the cold prospecting, I think it's a huge, huge value add when it comes to doing some of the research, finding companies, pulling useful data from there. It can then present that back to you. You can train it to do some of the outreach side. So you see on there, it says 5 email pass sequence. That's something I've trained Claude on in terms of how I want to go after prospects. For anyone that is into copywriting, PAS stands for Pain, Agitate, Solution. Keely Flood [00:11:31]: It's a specific copywriting framework. And so Claude has access to that already just because it does scrape the internet. But I trained it a bit further with my own writings, my own styles. So it learned, okay, this is how Keely writes his emails when it comes to the cold outreach side. So I essentially, and I'll show it here in a minute, I can go in, give it an account name, it can run, pull the data. And literally write the emails for me on my behalf. Now, going back to what I said at the start of this, I think it only gets you 80% of the way there. There's gonna be times where AI's gonna come up with outreach that you're gonna say, you know, I don't really love that. Keely Flood [00:12:04]: And you're still gonna have to go in and make some edits. But I think there's still value in using it because if you can get 80% of the way there and I can spend 20% of the time that it would usually take to write these things, just doing some edits and some tweaks. I bought back a lot of my time to focus on other things as well. The discovery call piece. This is what I was talking about before. After every discovery call that I conduct, my transcript gets uploaded. Claude can then go in, pull out all the necessary data, and send on my behalf something I call a game plan email, which is a very structured way of taking the information I learned, putting it into an email, and sending it off to the prospect for them to review and get back to me from there. And then the follow-up piece I talked about HubSpot with, and clearly you can do the same thing where we can have it set up where after the call it can set up follow-up tasks on your behalf to remind you, hey, you need to follow up with Keely one week from today, or hey, based on your call you just had, the 2 parties agreed to touch base on July 6th. Keely Flood [00:13:03]: And so Claude can then go and create those tasks, create those reminders on your behalf. So you, again, it's your second brain. You're not spending all of your time thinking what I should be doing. You're not spending a ton of your time in administrative stuff. You're spending more and more time just out there actually executing on your job, AKA having conversations with candidates and clients/prospects. This is— we're going to get into the live demo, so this is probably the part where hopefully questions will come through. I just want to show like very briefly what I just talked about a moment ago with the recap side of things, but how the process would work. The call ends. Keely Flood [00:13:35]: For me, how I have it set up I have all of my transcripts go into a particular G Drive folder. So when I tell Claude I need a recap email for my call with so-and-so, it immediately knows to go in there. It pulls a transcript. It then extracts those pain points slash next steps. It'll draft the email on my behalf. And then once I approve it, once I say, yes, this looks good, the last thing it does is updates it in HubSpot, updates the notes, sets the task, and then I can send the email out. I send the email out myself. When I say approve, I mean I approve the messaging and then I take it over to Gmail myself. Keely Flood [00:14:06]: I personally don't want— I don't want Claude sending anything, candidly. It's just me. It's my own 2 cents on things. I don't want it sending anything on my behalf. I'm good with it helping create it, but I still need to go and make my final edits and hit send myself. I just don't want to trust it to send things out because I've seen some horror stories of things that have gone a little bit haywire. So maybe I'm a little bit over-paranoid, but I'd rather take that one last manual step of taking the email, putting it in or taking the writings, copy and pasting it into my Gmail, making it, and then actually hitting send myself. So that's one thing I want to clarify with that. Keely Flood [00:14:36]: Cool. So I'm going to actually pull up Claude here, and hopefully we've got some questions as I kind of demonstrate this, because it should be fun to see this in action. I cheated, candidly, because it does take a little bit of time. So what you're seeing on your screen right now, this particular piece right here This is those skills, those agents I was talking about a few slides back, those 27 different skills I've created. This is one of them. And so what I did before this call, because it does typically take a couple minutes for it to do all the scanning side, finding all the data. And honestly, if anyone's willing to drop a prospect they wanna go after into the chat, I'll actually do a separate chat as I'm walking through this and have it run on their behalf. We can come back and look at it here in a few minutes, but it does take a few minutes to go through this stuff. Keely Flood [00:15:20]: Which I'm okay with because it's not me having to do it. And I can go grab my cup of coffee, come back, sit down just like that. It's now created for me a worksheet, which I, again, through prompting and teaching it, here's the things I want to see when you do these scannings of companies. It presented to me in a very nice clean format saying, here's some signals we observed. This company did go through an acquisition a few weeks ago, so that's why it says delisted from here. Here's the job they have open currently. It's even telling me here's probably the signals you want to lead with, a.k.a. this could be an angle that I'll use from an outreach perspective, as you see broken down here, here, and here. Keely Flood [00:15:58]: And then I have it set up from a tier perspective. This is probably a little bit overkill, but when I do have Claude go through and do some research, I will have it break things down by green, yellow, and red in terms of tiers. Green being obviously the best company to go after, yellow being, hey, could be good, and then red be like, why are we prospecting this account? That all this stuff is customized to who your ICP is and who you want to actually get in front of. Over working with Claude and teaching it, it'll get better and better too as well. So as you can see, it then basically gives me the snapshot and then it says, do you want to do the outreach sequence? So going back to again that workflow, and I don't want to jump around screens, so I'm trying to help everyone visualize this because I know when I've been on talks, people jump from one window to another, can be a little bit distracting. But if we go back to that one slide we talked about What it came to, how cold to close, it was doing the research, doing the scan part, and then doing the outreach, that past email sequence. We're at this stage right here. So just like that, Claude has now helped me target Mary, who's the COO, as you can see up here, of Certio, and it has written an outreach on my behalf that I can send out to her. Keely Flood [00:17:02]: Is it 100% perfect? No. Candidly looking at this right off the bat, I would have to go in and reteach Claude because there's something I did wrong. I hate em dashes. I never wanna see those. I think that's a dead tell for AI nowadays. So this would be something where I'd have to go in and actually fix this skill to make sure this never happens again. So there's a few tweaks you're gonna make here and there, but is it 80% of the way there? I would say yes. Like, I didn't know they closed the deal on June 16th. Keely Flood [00:17:25]: This could be new information for me. And then how it's broken down for here overall, I personally like this. This is my writing style. This is how I send out my cold emails. I like bullet points like this. I think it's pretty dang close. To being a finished product. I would then take this, go into my outreach tool, whether that is using within Crelate, whether it's doing just from Gmail, it doesn't really matter. Keely Flood [00:17:44]: Go in there, make the tweaks, and hit send. But as you can see, it goes a step further because I have it broken down into 5 steps: 1, 2, 3, 4, and 5. It broke it down for me because I taught it how to do that stuff. All right. I see a couple questions too. It's a good pause. What's your view on sharing candidate or client contact data with Claude? Is it saved? That's a good question around the data piece. I am always cautious about personal information, like going in there. Keely Flood [00:18:11]: That can be a little bit tricky because, for example, I think one of the best skills you can create is a candidate submittal package, like how you do your submittals. I think most people have like unique way of doing it and you can teach Claude how you want to do it, whether how you want to do the email write-up, how you want to do the submittal package. There is, of course, with that comes the risk of you're putting in candidate information, phone numbers and emails and things of that nature. So I'd be a little bit cautious around that piece. The part about does it save? Yes, it can save to that Claude MD file where it becomes like a running list on there. Obviously with a number of candidates you could be working with on a day-to-day and week-to-week basis, I think that's probably a little bit of overkill. That's why I think something like Crelate would be better off where all that data is saved within Crelate. And then I could always prompt Claude to go in and pull that data out from there on my behalf. Keely Flood [00:19:02]: So hopefully that answers that. Building these additional prompts to get the data you want, can you post those signals? I think it's best when thinking to yourself, how am I going to use Copilot? How am I going to buy back some of my time? One, start very small, meaning just start with one skill. Because for me, like if you're looking at my stuff, these are all skills I've created over the months of working with Claude. Like every month I'm adding new skills basically. I would say for most people, that's overkill. Don't need to start there. Just start with one skill. And then the question becomes, okay, Keely, what skill should I create first? And I always say, what is a process that you do almost daily, or at least weekly, you find yourself doing every week that can be taught to Claude? So again, the submittal package, I think, is one of the most obvious things to go to, to create first, because every week you're sending out candidates to clients. Keely Flood [00:19:54]: And so you can teach Claude how you do your submittal package so that way moving forward, I can simply go in and say, I want to do, like, literally you would do backslash candidate submission, can't submission like this because this is my own skill, can't submission on my candidate. And then you can put the person's name and it'll execute from there. Unfortunately, I don't want to put anyone's real name in here, so I'll do a live demo real quick as I pull up some slides and we'll come back to look at this here in a second. On my candidate, Eben. So that's going to run. I'm going to come back and show that in a second, but I'm gonna go back to the slides and walk through a few things because it does, again, it does take a few minutes. So I don't want us just staring at a Claude screen running through some stuff. I want to continue on and we'll go back and look at it here in a second. Keely Flood [00:20:33]: So here's like some examples and I kind of, I showed it with that surgical prospecting, but I wanted to again break it down of like other ways, spark some other ideas for everyone on this call of ways you can use Claude to start your day. Because again, I'm a big believer, start small, build from there. A very simple thing could be your morning brief, AKA I have it set up where it's connected to Crelate. It knows my pipeline. It can then, I go in and say what I need to get done today or cue my morning brief skill. Claude can then pull up and say, here are your tasks due today. Here's people you need to reach out to. Essentially, I look at it as, as my own VA, someone that can help me stay organized at times, be that second brain. Keely Flood [00:21:14]: So I'm spending my time working, sitting at that desk out there executing and not so much planning. I can have Claude do some of that stuff for me. The surgical account scan piece, that's what I showed a minute ago. You give Claude the name and then it can simply execute from there. When it comes to like Stephen's question around signals, for me, like candidly, I'll give a couple ones that I think are important. Obviously headcount growth, I think is an obvious one, 5% or more. It's like, okay, this company's growing. Chances are they're gonna need some hiring. Keely Flood [00:21:42]: any sort of major news-driven events, AKA they got bought out like Assertio did, they're expanding to a new site, they bought a new location, right? I was talking to a company last week and they were telling me about how they purchased a new office location in Dickson, Tennessee. Okay, if you bought a new office, guess what? Chances are you have to fill the thing. So that would be a signal if I could find that somewhere from scraping the web as well. I think also just overarching theme with this, When it comes to finding signals that can be useful for you, I always think of it in these terms. Whenever there's a change within a company, I always say changes present challenges, challenges create problems, and problems have a need for a solution. So whenever I see something occur within a company, some sort of change, I do that simple mental exercise. Okay, this change is going to result in what potential challenges? Okay, now this is a challenge. What could be then some of the problems that occur? Okay, now with those problems, am I the solution to said problem? If the answer is yes, great. Keely Flood [00:22:42]: It means it's time for me to reach out because there's an opportunity to help someone in some dire straits that's dealing with a problem that is eating up their time, giving them a ton of stress, whatever. However the angle I want to take reaching out to that prospect. But that very simple mental exercise, I think, helps to then realize all these changes could be potential signals for me to reach out to this person or to this company as a whole. Uh, follow-up question, if we connect Claude with Crelate's HubSpot, Claude can still access all the contact information? Yes, it can. Within Claude, there is the ability to limit what Claude cannot do within the tools it's connected to. To answer your question, yes, it has access to it, but there are some pre-built measures, which is one of the reasons why I like Cowork. There's some Prebuilt safety rails put in for when Claude connects to these things, whether it's by MCP, which stands for Model Context Protocol, or just like direct connect into Claude, which I'll show it because it'll probably help from a visual perspective. And I'm probably going a little bit deeper than I'd planned, but the question's asked, so I want to show everybody. Keely Flood [00:23:44]: There are this one about the the safety rails put into the connectors and via the MCP. There's certain things you can limit. Claude to be able to do. I have it turned on to a lot of things here. There are certain things that it, it needs approval every time where it'll say, hey, Keely, do you want me to update this HubSpot record? I have to say yes to that. I don't want to give it free range to do anything within there. There's also within this connectors piece too, on the tool side, almost every tool I've seen has it pre-built in there that Claude is unable to delete anything. So if you ever wanted to like purge some of your records, it simply cannot do it. Keely Flood [00:24:23]: Those safety rails were put in place, and this is again, get a little deeper, but back before Cowork came out, there's something called OpenClaw came out in December of last year or November of last year, and it was a similar tool, but it was open sourced and it was much more technical. And there were some horror stories of people giving OpenClaw access to particular databases and tools, and that OpenClaw wiped it completely out, like cleaned out their database. Deleted everything because it went crazy. Claude was created and these connectors were created to put some of those guardrails in so it doesn't happen. And that's part of the reason why when I saw Cowork come out, I thought, okay, I can do this. Like the OpenCLaw stuff, software engineer thing, like I'm like, I'm not that guy. I don't have the time or energy to learn all these things. But when Cowork came out and there's some of these guardrails and stuff, I thought, okay, this is something someone like me could pick up and learn from. Keely Flood [00:25:12]: I'm going to come back to this in a second just so I can stay on track when it comes to Some of the things that I've already showed as well, but just like the slides as well. We've already seen this in action in terms of I did a live demo on that piece as well. It does get smarter. Like I said, it can get smarter every week if you have it auto-scan, which is what I do. I have it set up Fridays at 5 PM where it does this weekly scan. It'll go back through all the histories, all the chats we've had since last Friday, pull out the relevant information, and then save it into that Claude MD file. So I'm not having to constantly update That thing over and over again. I will tell you this too, just from a learning perspective. Keely Flood [00:25:49]: If you're ever running scheduled tasks, AKA they run autonomously, like this one's set up to run Fridays at 5. If you find yourself creating quite a number of scheduled tasks, my personal recommendation to you would be, if you're able to do it, because I don't want to— I never really want to encourage anyone to like work on weekends. You spend time with your family and live life, so to speak. But I think it is useful if you have access to your Claude on the weekends is to have those scheduled tasks run on weekends, and then all you have to do is simply turn on your computer. The reason why I encourage that is because me, when I first started doing this, I had all my scheduled tasks run Monday at 8:00 AM. When I sat down, I thought, hey, why not sit down? I let these things run. I do my LinkedIn, I do my other things. I come back, I look at what's been done. Keely Flood [00:26:33]: What I found myself running into from an issue perspective was because the scheduled tasks were running, it was using up a very large chunk of my token usage for that time period. So if I ever wanted to use Claude for like something happened on Monday, I want to use Claude to do something different, I was hitting my ceiling pretty quickly. And so what I realized was I'm better off moving my scheduled tasks to occur on the weekends when I'm not sitting in front of Claude, when I'm not doing this stuff. I just let these things auto-run, and then when I sit down Monday, the outputs are already done. I can go and review and then go on with my day. Like a little tip for everyone, if you're gonna start using this, try to do your scheduled tasks if possible on weekends or at the end of the day at least. So that way during the time you're sitting in front of your desk, you're not using up all your token usage for when you might wanna use Claude in one-off situations. So I mean, ultimately like what changes? The whole goal of what I saw and the value I saw with Cowork is it gives me back my time. Keely Flood [00:27:28]: Like I am able to, draft up emails in a couple minutes. I'm able to do research in a few minutes. I'm able to do things that used to take me— writing a recap email, candidly, used to take me 15, 20, 25 minutes depending on like how deep I got into it, having to pull up the transcript, figure out all these things. Claude is able to run this thing on my behalf in a matter of minutes, and it frees me to do other things, focus on other parts. Like, candidly, a lot of times when I'm after like a coaching call, I will pull up my Claude. I'll have it run through my Post BD coaching process, and I go for a walk. I go for a ten-minute walk. I come. Keely Flood [00:28:02]: I sit back down. The email's been drafted up. The HubSpot contact can be updated. All these things are done, and I am able to just simply go on and execute from there. I say this part right. I don't think like this is incredibly difficult to do. I think there's a learning curve with it. I think the hard part is just the time commitment. Keely Flood [00:28:21]: Like it takes a long time to figure out. How to work these things, how to get the proper outputs, how to get the connectors, things like— everyone I think on this call is fully capable of doing it because as I was talking about, if I could do it, everyone else can. I'm a football coach. I was a history major. I have no technical background whatsoever. And I was able to get to this point just simply by trial and error, just spending the hours figuring out these things, making the connectors, setting things up, creating the skills, deleting skills, updating skills, all these things. You can do all these things. It's just a matter of putting the time in to figure this stuff out. Keely Flood [00:28:58]: And if you are willing to do that, I always tell people, start small. Start small, block off 30 minutes a day to say, I'm gonna try to get familiar with Claude Cowork. I'm gonna try to figure out how to use some of the stuff from my BD side of things. I'm gonna figure out how I'm gonna do my first skill. If you do the small little blocks, eventually you'll stack up and you'll get to the point like me, 6 months later, and you realize, oh wow, I've created all these things. But it was through like a slow, deliberate process I was able to do all that stuff. So we're gonna talk about how to build your own. I'm gonna try to walk through, 'cause I, you know, I think it'd be very helpful for everybody to build your own. Keely Flood [00:29:30]: And I will even, I'm gonna share a prompt here that will help you. Basically Claude will help you walk through creating your first skill. But going back to what I said in terms of like what you should build, I have it on here. Like think of one process you do almost every day, or at least every week, and then you can have Claude help you build that thing as well. And we're going to try to do that now. Could be a complete disaster or it could turn out really, really well. But what doesn't look like a disaster, at least I don't think so, is the canned submission package that I made up for Steven. I told Claude it was fake, so it went in and created all these things on my behalf. Keely Flood [00:30:01]: This is mine. I taught Claude how I wanted the submission package to look like. So you might look at this and say, hey, this isn't how I would do it. I was like, okay, that's great. You can teach Claude. One of the best ways to teach Claude is actually take your prior work history And feed it into it and start saying, here's my submittal package, here's my writing style, here's my LinkedIn content. The more data you give it about yourself, the better it can get to actually being not a perfect replication of yourself, but pretty close to what you would do. Because that is one thing that I think AI is phenomenal at. Keely Flood [00:30:34]: The more data, large datasets, the better it gets, the smarter it gets. And so what you see here on this right-hand side with the submittal package, This was created with me just dropping in. I'm like, here's my submittal, here's my submittal, here's my submittal. And over time, Claude's like, oh cool, this is how you want it done. You want to have this heading here, you want it to look like this, you want it broken down by first highlights, then first why they're good for this role, then here's what they said. Like, it does a good job of pulling all these things out and then helping you replicate it from there. And I have it connected directly to my Google Drive. So if I were to click this button, it'll take me into my G Drive and I'll be able to then make some edits to it if I wanted to. Keely Flood [00:31:05]: If it looks great, I can then simply download it right here. I'll do this. I'd show in the folder, open in Word, send it out, and just like that, what would have taken me a lot longer to create a similar package like this took a matter of— I mean, honestly, probably took 3 or 4 minutes. I know it's hidden behind those slides, but it took a matter of a couple minutes to get this thing created. All right, questions, anything so far before I start going into like trying to create our first skill? Okay, I think everyone just wants to see how you create a skill, so that makes sense. Okay, so a couple things here. I'm gonna jump around, I'll create the skill itself. When you go into Claude, Cowork specifically, and I think this last piece I didn't say this already at the start of the call, the plan I use, it's $100 a month because of the usage tokens you get. Keely Flood [00:31:46]: So I run out of my usage tokens pretty quick on the $20 plan, but I would encourage everyone when you're starting out with it, just do the Pro plan, the $20 a month. If you find yourself hitting your limit every single day, it may be worth the extra $80 a month. If you're not hitting it every day, stay on the $20 plan. I use Claude. Is that different than Cowork? Yes. So Claude, the browser version, right? You go on, you open up Chrome, you type in Claude, it brings you to— it looks like this, right? The chat. This is the chat right here. You go in, you ask questions, things of that nature. Keely Flood [00:32:17]: Claude Cowork lives on your computer. That system has access to the files in your folder. It creates files on your computer. It basically like, think of it like another app on your computer. Like you can't see my screen right now. But in the bottom right of my toolbar, Claude, the app is open right now. It's still Claude, but it's just, I would consider it the agent marketplace or the agent HubSpot. That's where you go to have Claude actually do things, create things, things of that nature as well. Keely Flood [00:32:42]: All right. I'm reopening Claude because for whatever reason it was not, and they just did another update this morning before I got on the call, which was a little bit frustrating because they do an update basically every other day right now. But for whatever reason, that update is making it a little bit— it's not letting me open Cowork at this exact moment. So other questions as I'm fidgeting with this thing, it's still giving me some issues. Alright. There we go. Took me a minute, but we got it. Alright. Keely Flood [00:33:02]: We're gonna do new task. Do this. Oh, that's what I wanna show too before I get in the new task. So back to this whole skill piece, if you're starting out, you can add a skill. If you do browse a skill like this, there's already some prebuilt skills within Claude that you can then simply download. So the MCP builder, that would be That's something to help you connect to tools if they don't have a direct connection link, but there's a bunch of other stuff in here as well. What I grabbed right off the bat was, and I'm gonna show you here in a second, it's at the very bottom, the skill creator right here. This is something that Claude themselves helped create. Keely Flood [00:33:35]: So now whenever I want to create a skill, it'll run through this task on my behalf to help me create the skill in real time. And we're going to do that because I'm going to grab this prompt and I'll share this with everybody too, by the way. But I just wanna show you like how in theory it's not as hard as you think to get this stuff up and going. Because if you're already prompting with Claude, if you're already doing some stuff with that, you're already probably 50 to 70% of the way there to creating your first skill. Because what I'm doing right now is I'm essentially giving it a prompt and then it's gonna ask me a few questions based off of this prompt that I gave it. I'm gonna answer a few questions and then it's go in and actually create the skill. So I want it to be a Word document. We'll do this. Keely Flood [00:34:13]: And then it's gonna ask me, you know, what roles do I recruit for? It knows me already to an extent because of that Claude MD file. So I'm gonna say I'll just do this one right now. And just like that, this thing's gonna start creating a skill on my behalf. And I wanna show this because, well, it's not gonna be perfect by any means because as you can see here, it's gonna ask me about my voice. Like I would recommend this is where you start uploading your own content, your own submittals, everything else, 'cause it'll learn from you. I just wanna show in real time like how simple it can be to create your first skill. It shouldn't be something that's overly top intimidating. It's something that anyone here can do and like leave today's call and start creating and be able to hopefully, right, buy back some of your time from a BD side, a recruitment side, however, wherever you're spending a lot of administrative time that can be offloaded onto Claude. Keely Flood [00:34:58]: This is where we can go to do that. As you can see here on the right-hand side, this is happening. This is popping up because it's— Claude is telling you hey, here's the skill we're doing to create said tasks. So on some of the other tasks that I did, it would have popped up a different skill over here. It's basically signaling it's using the right— telling me it's using the right things I should expect. That's why Alex said something too. Cowork's different because you can connect things. I mean, yeah, correct, right? Yeah, Cowork is different because it does connect. Keely Flood [00:35:24]: That's— again, Cowork is that like second brain. It can create things. The slides you saw today were created by Claude, a ton of like docs and the middles and all these other things. I don't wanna go through every single one of 'em, but like all that stuff was created with Claude. It created it in there. It's synced to my Microsoft Word, it's synced to my Google Drive. You can upload it and just like that, we're done from there. Other questions? This is kind of running through stuff. Keely Flood [00:35:46]: This is always my least favorite part about showing Claude is the moment where I have the thing running and it's like, all right, it's cool, but it's also like we're just sitting around, which is always a little bit unfortunate for my part of things. Kortney Harmon [00:35:57]: Really? I'm gonna chime in 'cause I use Claude Cowork on a daily. Like it is my brain. Most days it's my brain. But like you just said, submittals and like judging like what you say, like you can have it go look in your Outlook and say, get the last 10 submittal emails that I have sent, consolidate what my language is. I got an email yesterday, don't tell anyone at Crelate, and I'm like, did I do that? I was like, literally scour my email, see if there was a, like where this was, what happened, where was the call to action. It can help you, like instead of you searching for hours in your Outlook to Like, what happened? Where did that go? How did that happen? It can give you stuff so real time. I love it. It has been a lifesaver for me. Keely Flood [00:36:39]: Yeah, right. It's like, it's just such a, like, a time saver. I would be very sad if I had to go back to doing some things without Claude. I'll just say that. Like, I was like, well, how far do I want to get? I would be a little bit upset if I had to go back and do some things without it, because again, I'm able to get so much time in my day back that would have taken me a lot longer. I mean, yesterday I was on 5 different calls. And like, if you think about it, if you have 5 calls in a day, which most of us should be shooting for, whether it's candidates or clients, you got the recap, you got to update your CRM, you got to do all these things. Like each call in and of itself creates roughly 20 to 40 minutes worth of administrative work after that point. Keely Flood [00:37:14]: And so if I have 5 calls, we're talking anywhere between 100 to 200 minutes of admin time, which I, again, it's important. It needs to be done. You should be logging things within your CRM. It is your second brain. Like all these things have to occur. But if I can have AI handle that piece and I can do other parts of my job, that's how I'm able to, you know, the big phrase of like 10x-ing yourself, but that is how you're able to make yourself more superhuman because AI's handling a lot of things that you don't have to handle anymore that are necessary though. Like I have coaching sessions agenda. I'm trying to see if I want to pull up a specific one. Keely Flood [00:37:48]: Like here'd be one. I don't think they would care that much. Like, so here's like a client I met with and here's the recap email created for them on my behalf. Just like that, all that's done. It was an hour-long call. There's a lot for us to talk through. It captured everything, put it in an email. I also, with everyone I work with, I help them create a sales playbook. Keely Flood [00:38:04]: So it takes all the data, all the things we talked about on that particular call and presents a nice clean Google Doc for me to say, yes, this is what we talked about. Again, here's how we're gonna create trigger events, things we're gonna look for. Here's the mental model I spoke about. All this stuff gets created within Claude. And actually, because I didn't actually do this, which is good, it's then, as you see over here on the right-hand side too, just to further go deeper in this whole Claude thing, it will walk you through its progress step by step. And now I said it's good. It's going to go in and update HubSpot with those particular, those 2 particular contacts. And then it's able to create the Gmail draft, which I already sent out. Keely Flood [00:38:40]: And we're good there as well. And as I'll go back to the skill too, because there's something else I wanted to show that I think most people would find kind of cool. Okay. So this is still running. Because I did cheat, because that's what you're supposed to do for these talks. I did cheat a little bit and I came prepared to show some particular things because if anyone noticed this side here, this research piece, this autorun I was talking about, I want to show you what that looks like when it's all said and done. So again, like hopefully, and I'll take a pause before I go a little bit deeper. My intention today was not to like overload everybody with like, oh my gosh, like I always tell people and I didn't say at the start of this call, so it's bad on me. Keely Flood [00:39:15]: I always say if you can leave from all these talks, there's a lot of great speakers that are happening right before me and happening after me as well. If you can leave every talk with 1 to 3 actionable items, it's a great talk. If you try to do everything you see on a talk, you're going to hit a wall and you're going to ultimately do nothing because it's going to be intimidating. It's going to be too much. And so what you're seeing on here is I'm jumping around from this to that and the other. If I'm me 6 months ago, And I was presenting to me and I saw all this stuff, I think my brain would explode. I'd be like, oh my God, this is a lot of stuff. Like, I don't even know where to begin. Keely Flood [00:39:48]: And that's why I keep saying, like, start small, start with a single skill. The stuff you're seeing now, this is months of me doing this stuff and it's still not perfect. There's still em dashes in my things. There's still things that need to be fixed. So if you can just take 1 to 3 things away from this stuff, from these talks, you'll be better at your job. You'll be better for your recruiting firms. Everything will be better. if you start small and go from there. Keely Flood [00:40:09]: But what I want to show you with this piece, and this is what I mean about the HubSpot piece too. So going back to that whole scheduled piece, I have it run every Monday. It finds 5 new companies for me, then pulls out the signals that you can see here on the right-hand side, and then it updates my HubSpot, which again, I cheated. So I had this stuff ready to go. Oh, there's Kortney, 'cause you emailed me. I don't wanna show that one 'cause I wanna show you more information. I'll just open it up here. This should open up. Keely Flood [00:40:30]: And as you'll see here in a second, So going back, I'm jumping around a little bit, but Tru Software is what I identified. They're hiring. And then Claude went in and created Tru in my CRM for me. It added the people for me on my behalf as well, and it even included description down here as well. All that stuff was done, and you know who did it? Any— who did zero of it? This guy. I didn't do a single thing when it came to doing the research piece, but I put in the time to train Claude and get it up to speed. So it can then go and do these things on my behalf, and I can then simply sit down, get caught up to speed, and start executing on my day-to-day plan. If anyone's getting Bing locked, Claude was asking me a lot of questions. Keely Flood [00:41:11]: Sometimes it can be a little bit annoying with like some of the questions it asks. But so here you go. Like, this is all approved. As you see, like, it's done. It approved it. Everything's taken care of. The HubSpot was connected. I want to show you that one, but I want to go back to this one because that's the main thing. Keely Flood [00:41:22]: It's running through this test of creating the skill. And before you know it, it's gonna be done. And now we'll have a submittal skill we can use from now until the end of time, in theory. Again, it doesn't have my voice, doesn't have my content, doesn't have how I do things, but you're seeing in real time how to go about creating a skill from start to finish with like a few prompts, a few inputs in terms of how you do things, how you write and all that stuff. And Claude's able to create it from there. So I'm gonna go as that's running, and this is like, again, this would be like a, This is a cool little graphic that, guess what? Didn't create it myself. Had Claude help me do it because I'm not a graphic designer by any means. I should take all my stuff and try to put it in a nice, clean looking PDF for everyone to look at. Keely Flood [00:42:02]: And this is what I created as well. But as it says there, if you got questions, like ask away. That's what I'd say. If you're ready to build your second brain, like let's talk. You can connect with me on LinkedIn. I do have a summer cohort coming up and I do work with people one-on-one. I want to just quickly plug that, but I want to go back into now say, what questions do you have for me? Because hopefully, hopefully this has like been beneficial and pull this up as this is running. Hopefully this has kind of like blown a few minds. Keely Flood [00:42:24]: I hope, I mean, you can now kind of realize there's a lot of really cool things I could do with Cowork. And ChatGPT's come up with its own agents as well, because I typically get this question, hey, Keely, like, how do you know? What if ChatGPT's model is like better? Like it comes out and they say it's better. I personally, and again, this is my own personal stance, I think the LLMs themselves, Gemini, Claude, OpenAI's ChatGPT, all these things, they're all in a race to improve their models. And so if OpenAI's model is now 5% smarter than Anthropic's, which is Claude, there's a very good chance Claude is going to try to catch up to it because they're competing. So the way I look at it is I think there's more of an opportunity cost to switch from one model to the other because one is supposedly better, that it's just not worth that opportunity cost. I would rather get really, really good at one particular model set things up like I have it set up here and ultimately just live and breathe and die in that particular model than chase the next shiny object. Because having to transfer the stuff over and having learned how to do this, having to do the connectors and all these other backend work, I personally don't think it's worth the time investment doing that. So questions on any questions at all? And this is, this unfortunately, I'll see if I can open up on another screen here. Keely Flood [00:43:35]: It created as an HTML file. I'm going to tell it, I'm just going to say it looks good. Part of that skill creator, it's within that skill creator, it is programmed to ask for your feedback before it finalizes. Then that's kind of what's happening here. It's basically saying, hey, look, here's how if you use a skill, here's the output. Here's if you didn't use the skill, here's the output. For the sake of like time and simplicity, I'm just going to say it looks great. It's going to then create the skill and here you go. Keely Flood [00:44:01]: We created a skill in the 12 minutes I sat here talking. Claude helped me create my first submittal skill. And how do we do it? Again, not very hard. I dropped the prompt in there. It will ask me a few questions. Again, as I had in that slide, how— have Claude help you build within Claude. It asked me a few questions to answer and upload and refine it. And then from there, it simply executed. Keely Flood [00:44:24]: The hard part then becomes once you go through there, you're gonna have to make some tweaks. You're gonna look at this. There might be some things you like, don't like. But overall, I would say it's probably 70%, 80% of the way there of a completely finalized skill. And if I don't like it, I can go into this piece here because this is what the skill is, make some edits, rerun it, and once it's finalized, that skill's with me for life. Which connectors do I use daily? I use G Drive just because that's where all of my transcripts go. I use a note-taker tool called TLDV. I've been using it now for a couple of years. Keely Flood [00:44:57]: And I had it set up before Cowork came out. And this is a long story, but it'll make sense. I had it set up a long time ago where TLDV transcripts with Zapier would go into my G Drive folders. And then when Cowork came out, I just said, okay, Cowork, you have access to that transcript folder now. And so it's able to pull all the data from there. So I use that. I use, um, obviously HubSpot. That's where I can save all my coaching notes, my prospects, everything else from there. Keely Flood [00:45:22]: And then I even have Canva connected to it as well. And then your data tool. So if you use like, again, I have ocean.io, there's Apollo. If you have ZoomInfo, I think those are like pretty much 3 non-negotiable, like you want to have them connected would be to your emails, to your email provider, to the data and to your CRM. If you have those 3, those are 3 powerful connections. The rest of the stuff, like every day there's more connectors coming out. And if you're using those tools, it's awesome. I'd highly recommend connecting to it. Keely Flood [00:45:50]: But I think bare minimum, if you have those 3, Claude Cowork can like supercharge itself because it has access to all these tools and do some really cool things on your behalf. And this would be like for anyone that's going to ask some of the connectors here, when I say like it's adding, like here's stuff I have, but every day this list is getting longer. Like, and there's stuff on here, I have no clue what it is. There's so many things. And I think every tool out there, I'm sure their tech team, if they haven't already, The leadership and the tech team, everyone's working on how do we make this connect into Claude? How do we make this connect directly in the chat? Because if it does, you can do some cool stuff with it. And I think from a customer and user standpoint, everyone who uses a particular tool is probably asking their customer success manager, can this connect to Claude? Can this connect to Claude? And they're all trying to make sure it does. If not today, it will soon. All right, there's that. Keely Flood [00:46:41]: So yeah, there's a skill. I'm gonna hit stop sharing and maybe answer questions there because that's about it. Kortney Harmon [00:46:46]: And to add on to what Keely was saying, just like ChatGPT and all the other LLMs and everything that's out there, the tools that you're using, just like Crelate and everything else, they are changing by the day. It used to be quarterly pushes and content or like what was changing in your products. And now they're weekly pushes and daily micro pushes. So I would also encourage you to really understand what your tools that you use today are. Because they're developing AI. Um, I'll put it in the chat again, uh, our free trial. And if you actually reference FDE at all, or this conference, you'll get 20% off of our AI agents. But AI agents and skills and everything, they're literally every single place that you're using stuff. Kortney Harmon [00:47:28]: My note-taker, I use Otter, and it like gives me a summary. I can ask it questions. It has its own LLM built in as well. Keely Flood [00:47:35]: Yeah, it's cool. It's accelerating quickly. And I think again, for myself too, I sit there some days like, how am I ever going to keep up with this stuff? But I just remind myself, what's one small thing I could help— AI could help me do? Because thankfully, at least I believe this, in the recruitment industry, no one's changing jobs or no one's working with a firm because they talked to an AI chatbot. I've made this joke before and I've said maybe in 20 years, like my kids, my eldest is 10, maybe when she gets in the workforce 15 years, 12 years, whatever it is from now, She might feel comfortable because she's grown up around AI to say, okay, I need to find a new job. And AI's like, great, here's some places you can apply. Here's the interview process. Like, AI handles it all. I think for everyone else, at least in today's day and age, no one's trusting one of the biggest decisions of their life, AKA, am I gonna change jobs, to an AI chatbot? So that means me, the human, is still an integral part in this whole process. Keely Flood [00:48:25]: And that's what allows me still to sleep at night with some of these things. Like, AI can do all this cool stuff. It can buy back time to do all administrative stuff. But when rubber hits the road, right, when someone has to make a tough decision, do I take this job? Do I make this change? Do I hire with this? Do I work with this recruiting firm? You still need me, the person, to be part of that process. And that to me is reassuring. Kortney Harmon [00:48:43]: Yeah. Question, is Crelate MCP ready? It is. That is something that is actually actively being worked on and being pushed right now and worked on in sandbox environments. So yes, absolutely. I believe we are ready. I, there was just a webinar this week about timeline. I can get back to you on what exactly that timeline looks like for those of you that are interested, but it is actively in conversations. So yes. Keely Flood [00:49:07]: Are you going to do a webinar showing people out? Because I think it'd be awesome if you like, I'm sure you have a team setting up, like show them out. Cause it's, I think for most people, when you see MCP, it's like for most people, cause I know with the MCP, I at first was a little bit intimidated because there's API keys and all these other things. Kortney Harmon [00:49:20]: Webhooks, the whole thing. Keely Flood [00:49:21]: Yeah. It's like a lot of like, I'm like, uh, so I'm sitting there with YouTube University trying to figure these things out, but once it's set up, oh my gosh. Life is just better. Kortney Harmon [00:49:30]: Yep. I've gotten to learn way more about this, obviously working at a software company between MCPs, APIs, and webhooks. I'm just like, it's like all the acronyms. It's like, which one is it? Um, but yeah, it's been, it's been definitely been a learning curve for sure. But yes, we will be doing that. We did it internally for our clients today, this week, and we will be doing it externally for prospects that are interested. But if you are interested, Reach out. I don't know if you're a customer today, but if you are not, shoot me a message and let's get a demo on the books so we can talk through that with our VP of sales. Kortney Harmon [00:50:04]: So anyone else have any questions? I know Keely put his LinkedIn profile up here as well. If you are not connected with him, please make sure you connect with him. Amazing content, amazing stuff all the time. Keely, thank you so much for your time and— Keely Flood [00:50:19]: Thank you. Kortney Harmon [00:50:21]: Real-world, applicable, hands-on, uncluttered co-work live on the spot. I applaud you because mine does not. It hates me some days when I do that. So yeah. Keely Flood [00:50:29]: Again, I was a little bit worried for a minute where I couldn't even get to the co-work tab. Like, this could be going really horrible. But again, it's not perfect. So I love to show the non-dressed up, everything works really well. It's like some things don't work really well. Like, there's some dashes still. Like, it's still not perfect, but it can get you back some time. So I appreciate having me on. Keely Flood [00:50:44]: Thank you everybody for the questions. And yes, if you haven't connected with me on LinkedIn, please do. Love to stay connected with everyone. So thank you for the support is what I'd say to that. Kortney Harmon [00:50:52]: Well, thank you, Keely. Have a wonderful day. Thanks for listening to this special FDE+ virtual event. Do you want to be a part of the conversation next time? Our live sessions are free, open to the public, and packed with timely insights from some of the brightest voices in our industry. To follow Follow the podcast, stay connected with Crelate so you don't miss the next event. Until then, keep learning, keep evolving, and we'll see you at the next one.