92. How to Really Use AI in Your GTM Team

Hello and welcome to another episode of the Operations Room, a podcast for COOs. I am Brandon Mencinga joined by Bethany Ayers. How are things going this morning? Back in the first week of January, back at work.

It's so hard. I wish I could be like, oh, I'm cheery and I'm great and everything is awesome, but that week back, I don't know, hopefully it's not going to be all of January and it's just the first week, but I'm tired. It doesn't help that I went to a dinner last night. I clearly have no stamina anymore.

So I'm physically tired. I'm emotionally exhausted and not my normal bubbly self. How about you? You know, I just realized, I was thinking about this this morning, like the actual new year is in spring, to be honest, because spring is when, you know, you get more daylight, it feels better.

There's like a spring in your step. You're thinking about the future, you're optimistic. The new year is not January 1st, because it turns January 1, we're forced by the calendar and suddenly we're supposed to feel like, yeah, it's a new year. It's a New Me.

It's resolutions and it is deadly cold outside. It is dark. You're being forced back into work off your comfy couch and it doesn't feel like there's a spring in your step. Yeah, I 100% agree.

And so the one good thing is that we've had the solstice and the sun is, the days are lengthening from here on out, but you don't notice the difference until February. January is just dark and grim, and we've been pushing hard since the sun disappeared sometime in November. So what's happening on the work front? Anything noteworthy?

The really big news is I have a new sales leader and new marketing leader starting next week. Okay, that's big news. Cannot wait. Sales and marketing, both starting in the same week.

Okay, well that's actually pretty ideal. Well, they were supposed to start on the same day, but because of calendaring issues, we have one starting Monday, one starting Tuesday. So yeah, that must be thrilling. You're like, oh my God.

It really is. And also, there were long notice periods and so there's been lots of pre-onboarding so it's not like they're going to have to spend months learning. They probably are not thanking me for the Google Drive that I created for them as their Christmas reading. But do you have Gemini?

So Gemini, I find, does research that's like dealing with a partner analyst, everything Gemini produces is super dense. Has no personality, but I kind of have like faith in the amount of research. But Nano Banana now does infographics. And so I'll just take the really dense information and then ask it to produce an infographic and they are fantastic.

And so for my go-to-market leaders, they have all of the like super dense information and then they have five infographics and I suspect all they need to do is look at the infographics, and that sorts everything out. But Nano banana was having some sort of massive issues. So each infographic took about three days to produce. And it just left it and it looked like it was still producing.

It's like, well, let's see. And I would check in every day and then on the third day that arrived. That's crazy. So the compute time, sorry, I'm just like trying to think this route.

So the computer time to create your infographic took three days. Is that what you're telling me? Yeah, but luckily I could run all five at the same time. Wow, I'm just imagining the compute center and Albuquerque cranking away, consuming vast amounts of power to power your infographic.

Well, but I mean, clearly it's not my infographic that's caught, you know, I'm in a big queue and I'm getting one, one billionth of the compute power, but it's taking three days to do it, but highly recommend infographics. So I am super excited for the offsite that is coming up next week for us. So we have our plan of action. We've got day two as you kind of do these things.

So day one is all around the company and our strategy and alignment and buy-in and all that jazz. Excitement energized for the new year. Day two is all about the AI upscaling. So we've got three tracks in the morning.

We've Got our friend Charlie Cowan coming in to facilitate the entire day, but also doing his bit on chat GPT in the morning for his track. We've got Johnny Ball coming in for lovable, for non-developers, which I think is going to be super awesome for the company. And then this other fellow named David, who's much more of like a real developer of developers. He's actually CEO of his company, but he is a developer at heart and he's very ingrained in getting the best out of AI for the development side in his company.

So he's coming in too. Do the cursor training in this case, do that three tracks, and then in the afternoon we have six challenges. And we're going to have people self-select into the tracks in the morning, self- select into the challenges, kind of normalize the group sizes for the challenges in the afternoon, and go at it for three hours, produce a couple of demos of interest that we can present back to the company, and you know, fingers crossed it works well. I have a lot of confidence in Charlie to kind of facilitate the arc of the day to make it all sensible and kind of drive it forward.

And he has just so much energy and enthusiasm that he gets everybody interested. It's amazing if you just think about the energy givers versus the energy takers, and as leaders, we need to be energy giver a lot. And sometimes I naturally have that energy giving and sometimes I don't like this week, whereas Charlie seems to just always be able to enthuse others and give energy. That's really authentic and also a great communicator can really simplify the messages in a way that's really I think for an audience of varying degrees of skill I guess when it comes to AI you can really kind of hit.

Yeah, so our training we're actually moving away from chat GPT and focusing on Claude skills and so learning how to build automations and better outcomes via Claude. So that's what our training is about. Yeah, I feel like we're one step behind you in this respect. I suspect in our next go round, it might be something similar.

I think the thing I'm most excited about is the lovable non-developer track because I think for 2026, I think in earnest with our CTO behind it as well, we want to put rest of the company in a position to ultimately develop. Actual code that gets injected into the product, obviously, with quality oversight from actual real developers before anything goes into production code, obviously. But I think that ability is going to be phenomenal. In particular, we've got one person that's setting up our new propositions team where she's kind of going early into markets that we're not currently in, and her ability to work with those individuals, those prospects and customers, and develop some prototypes and try to flex the muscle of what actually is compelling to these people and do that using Lovewell as an example, I think is going be a super obvious entry point for an individual that is not a developer to create real value in the company.

So she's super excited, I'm super excited. I think it's going to be fabulous. We'll see how it plays out. We've got a great topic, which is how to really use AI in your GTM team.

We have an amazing guest for this in Donna McCurley. She is the creator of the AI sales operating system and a go-to-market advisor for mid-market SaaS teams, turning Microsoft Copilot 365 into revenue engines. So Donna had talked about someone needing to be the AI overseer and she had pointed to rev ops as part of that. In your view, what does governance actually look like in practice?

So agents don't become this shadow system. So there's a lot in governance that we should be looking at. So one level is, and this is where I'm going to talk about Matomic, is what data do you have and who has access to that data, and who slash what has access to that date, so humans and agents. And so you need to do a certain amount of data categorization and identify what is sensitive information that should not be fed into your models and what is information that's safe to go into them.

Or be used by your agents. And somebody like us can help you with that. Then you also need to identify and decide what your policy is on shadow AI or not. And depending on your risk appetite, you can completely lock it down, have whitelists and blacklists of what URLs people are able to access and get quite draconian with it.

You can decide. The reason why people are using shadow AI is the AI tools. You're either not providing tools, which at this point, if you're not providing them, I think you're in trouble, or the tools that you are providing are just subpar. That would mostly be co-pilot, but it just doesn't work as well because it's so locked down compared to Gemini, Claude, and ChatGPT.

So again, your risk appetite, maybe need to loosen it up a bit, but make sure if your data is in a good place. You can loosen up your controls, bring in something that's not Microsoft to help you. And then you can start to think about agents. And by agents on the desktop, I mean like clients, not AI agents.

So agents that recognize what's going on and what your employees are doing. Employee spyware, which is a very popular in bigger companies in startups and scalps. We don't have to tend to deal with it. Or you can do something more lightweight like browser plugins that just scan.

All of the text boxes that you have that your employees are putting text into before they do it and either block it or just alert that there are employees that are behaving against policy and then you can address it with behavior change. That's with the CSO hat on of like super locking it down. I'm guessing our audience isn't worrying as much about that. They're fighting their so to loosen things up, which is then around policy and trust and balancing.

Data security risk with falling behind and not using AI at all, an existential risk, and trying to decide which way you need to go. Yeah, that makes sense. And then do you have an opinion on this revops ownership thing? So there's a fleet of agents within the company.

I think the rev-ops should be responsible for rev-op. It's agents but not others. One of the challenges with agents going forward is that there's a lot of cross-functional work and so I suspect over time there's going to be some sort of AI agent office that's looking at it across the board rather than specifically within rev-ups. A lot what Donna was talking about is.

Specifically within the revenue function. And so then it makes sense for rev ops to look at it, but there's gonna be a point where it starts to touch finance quite a bit. And then it's beyond rev ops control. And anytime you're actually talking about cash collection and cash reporting, then I think somebody in finance needs to own it.

So the other bit that she had spoken about was this opinion that on the AI-SDR front that the split that she sees being the optimal split between AI doing a lot of work on behalf of the SDR is a 70-30 split whereby 70% is agent automation and 30% is the SGR doing the human in the loop kind of like. Points of qualification of the AI agent outputs, I suppose, in this case, and also to some extent still doing the live calls with Outbound, what's your take on that 70-30 split? Yes, we ended up talking a bit around what is the role of an SDR and what are the new skills that you're looking for when hiring SDRs.

And I do think this is one of the roles that's changing most rapidly, where the people that you are looking for are way more systems thinkers. They're systems thinkers who are motivated by money is I'd say the profile that youre looking for. It's almost like an ops person that you're Looking for for SDR is now because it's like somebody who plus being motivated money. Likes to problem solve, and this problem you're solving is how to bring in good prospects, cares about the problem because they like to earn money, and are systems thinking and curious around technology.

It's not about the resilience of hitting the phones and cold calling anymore, although motivated by money means that you're probably willing to pick up the phone. I completely agree, but there's still, I think, isn't there like a live outbound call by a human to another human that still needs to occur? It always swings back and forth, doesn't it? Because you'll end up with so many cold calls.

Everybody screens it. You can't get through to anybody. Nobody answers. And then everybody gives up on cold calls, so then the numbers go down.

So then when you start doing it again, people answer. So it kind of depends where are you in that pendulum as to whether or not cold calls are effective. What might be more effective for right now, I don't know, is, I mean, how many thousands of emails do you get a day now, Brandon? There are just so many.

And so many of them are. Bad, like wrong company. I got one the other day that was like, we're so impressed by what Matomic's doing in the healthcare monitoring space or something totally random and went on and on and on. And this is like, and then wrong names, wrong companies.

But then through all of that mess, I got a really good one that I haven't talked to a human yet, but are following through in a process. Yeah, and this may be where this operator precision that you're talking about comes in mostly. Yeah. And also like, what are you offering in your outbound now?

So this one was LinkedIn and how to use LinkedIn more effectively for sales processes. Had noticed a couple of mistakes that we were making. Can you send a Loom video explaining the mistakes? Okay.

Yeah. We invest quite a bit on LinkedIn. We're not, you know, we can always get better results. And then I forwarded it around to the company of like, this is a good email.

But I'm afraid to ask for the Loom video because I don't know what's going to happen and what kind of like aggressive sales I'm opening myself up to. And then I was like, well, you know what? Why should I be afraid? I'll just ask for a Loom Video.

So now he sent it to me. I have yet to watch it because I have no time, but it was like watch the video and then let me know if you're interested in chatting more. So super lightweight, not an aggressive push, but giving me value from the very first interaction. Smart prospecting is going to make more of a difference, whether it's on the phone or not.

But also you might need to do it on the phone because thousands of emails every day. I've been in sales enablement for a little over 20 years, and as a lot of us know, sales enablement organizations typically have to run pretty lean. So being resourceful and figuring out how to support an entire sales organization is pretty critical. You've got account managers, you've got, from an AE perspective, you got enterprise, commercial, S&B, and then you've BDR.

So everybody has a whole lot of different things that they need to do. And so my initial thoughts were, how do I make sure that the sales enablement organization can support everyone? And when AI came along, it was like a dream come true for me. Because when I think about AI, I think of building agents as if I were hiring another person in my department, who will show up every single day, they're never sick, they do exactly what they're told, and they worked 24-7.

As long as I teach them how to do this. So I built what I refer to as AISOS, which is simply an AI sales operating system. And at a high level, what it does is when you look across sales activities, or your sales process, what is does is it does exactly what task a seller needs to do. So when you, hey, I'm in, you know, stage one, and I'm doing, let's say I need to do discovery.

What are all the things that a seller needs to do during that? You know, they may have to research the account. They may need to research The Buying Committee. They may to scrape for buying signals.

All of this now is something that an AI agent can do. Whereas in the past, this could take a lot of hours for sellers to do. We know that sellers like to skip steps and so, you know, it's like, oh I can wing it Well now a seller can show up for that discovery call completely prepared with all that account information buying committee buying Signals and even outreach messages that got them to that place and that's why I developed it because I thought how do I make Sure, I'm supporting the sales organization so that they're able to convert and feel confident and comfortable showing up The second thing is we can look across a lot of sales organization and a lot of sales leaders have been promoted into a sales manager because they were good as a seller themselves, but it doesn't necessarily mean that they've been trained on what it means to be a real sales leader.

And so you see these sellers who want to perform at an A level, but they may not have a sales leader who has the competencies to help them with maybe a skill that they're struggling with or product knowledge. And so they're at a disadvantage when that happens. And then also sales leaders are spread really thin. They may have eight sellers that they are trying to support.

And it's more like bring your deal to me, yes, yes yes, go or not really having that opportunity to coach. So AISOS does that. It is your 24-7, very highly competent, PhD level sales coach. So a seller can go in and say like I'm struggling with being able to you know This client has went dark You know I don't know what to do now and they can actually type in those questions and the AIS OS will provide them with hey Here's five ideas like which one resonates with you and actually build out like let's say I had a seller the other day This was their exact situation and you know it provided them with the reverse timeline Hey, in the discovery call, your client, Joe, said that this was their timeline they were trying to make.

Let's reverse engineer that and provide them with the timeline and, hey, Joe. If we're still going to try to make that November timeline, here's how this is going to look. We probably need to move forward. He was able to get Joe back on that joke, back on a call.

So it's things like that that his sales manager probably wouldn't have been able to have done. We tend to be very practical in the podcast. And what I love about your content is you're building all of this. You're not buying a bunch of tools.

Like, can we talk a bit around how are you actually doing it? The post that I got really interested in was when you took all of your top sellers recordings, analyze them and discovered what was actually why you were actually winning. So rather than answer, maybe that one is like. How to get started for somebody who doesn't have an A.

I. S.O.S.

And doesn't have a Donna but wants to have that in their organization, what would you suggest? This is how I got started. Is I went into CoPilot, and when you go into Copilot, there is a AI agent, like it's really, really intuitive. When you go in to the AI agent I believe that there are three things that you need to do.

The first thing that you to do is you have to decide what output you want that agent to do, it's just as simple as, you know, I want it to write an email, like you are my Ellie email. And so understanding what you're trying to achieve with that agent is first and foremost. The second thing you do is you have to give it instructions. And we all do this when we hire people.

We give them instructions. Hey, here's how you write an email. Now. I actually use a framework to help me rinse and repeat my instructions over and over again, which I like to teach.

And so being able to write instructions is the second thing, and it is simply, hey, how do you do the task that you assign? And then the third part of that is the knowledge base. And the knowledge base is you think of it as best practices. So what are some really good emails or campaigns or writing instructions that you have and you take that and you put it in the knowledge base.

That's it. Like you're done. You've given it an output, you've given an instructions and you've given it a knowledge base, now go play with it. Now go iterate and see what did that produce?

What was the outcome? Ooh, how can I go back and better improve my instructions because I want it to remove maybe the hope you're doing well kind of stuff or I want to use a specific email framework, things like that, and that's the best way to get started. So, I have a couple of questions. One is the first way of getting started is basically thinking about how your AI employee can do stuff.

So write emails for you, help you unlock a deal, the person that you go to and say, do this specific task. But what I always feel like is a really big unlock is when it gets starts to be tied in with automation. So when you're talking about the beginning, the employee that works 24 hours a day, so your SDR who's prospecting all day. And I'm really struggling, and I've been struggling all year, no matter who I speak to, about this AI automation true agent and how to do that.

Okay, I love that you asked that because here's where I like to say is this is my kind of soapbox is I feel like everybody's trying to get to ninth grade way too fast. So what I like to say, is that if you have got to build out your agent to advance them from first grade to second grade to third, you need to get your agent, one agent performing really well before you ever start to automate it. And so once you have your agents automating well, then let's say the next part of this conversation in a workshop, if we were doing a workshop together, I would say, hey, let's get your agents that can offload non-revenue producing work, the research, the signal scraping, all that.

Let's get those built. Now let's talk about how to run a trigger. Let's say we built four agents in our first workshop. Now let's take those four agents and now let's talk about how we could automate them Like how do we create that trigger to put them to just kind of run on their own and I wake up Monday morning And they're there.

Okay, so let's a real example here You're let's say you're an organization and my first question in this workshop too before we build out your you know Your auditions is thinking about okay. What are your growth levers? And so most organizations they will say like hey I have an operational efficiency that I need to pull, and I have a land and expand for my account managers. Okay, let's separate those workshops.

Let's only talk about operational efficiency. Let's not talk about operational efficiency because, again, I'm going to be a little bit selfish today on what we're looking for in our organization, and that is pipeline generation. So new business pipeline generation, that's a part of it. Operational efficiency is going to be a part.

I think you'll like this. Awesome. Okay. So I was just trying to figure out.

So yeah, I guess if I just say the number one unlock for us is pipeline generation. If you look at that as the most important thing. How should we get started? Ooh, okay, wait.

Let me pump the brakes. So you're looking more at your lead generation than you are more about the efficiency of your pipeline. Yes, so we don't have enough pipeline to worry about how efficient it is. We just want more in the top and then we can worry about the efficiency later on.

Okay, well, because once we get it in, we win 40% of it. And so but I think that's because customers have to work really hard to find us. And by the time they find us, they want to buy us. So I want to put more companies that are interested.

So even if our win rate goes down, we at least know that we're talking to more and kind of planting seeds for the future of other companies who would like to work with us. So your agents, what comes top of mind for me is, your agents would be, hey, if we were to take a seed list, a list of companies that have done business with us in the past and have had success, and then we wanna find more of those. So the first agent I would build is gonna be that look-alike agent, where we're taking that seed list putting it in, saying, hey what's not in our CRM, and being able, hey now go out and find more companies that are like this.

After that, now that you have, let's just say you're a targeted. New opportunity list so the next thing is because you guys understand so much historically about these companies in the past is taking those calls and doing kind of what my post said taking those. Calls and extracting buyer intelligence from those is first and foremost so now we have an an agent that is finding lookalikes and we have looking at our past we have agent that's going to find, you know, buyer insights.

And so those buyer insights is I want to know who are the people that are on these calls? What questions are they asking? What objections are they running an AI agent to look for those types of patterns is a goldmine. And so when we go back to writing the instructions, you're saying, hey, I'm going to load, you knows, calls from let's just take a month, you don't from last month.

I'm going to load those in there, and I want you to tell me who was on the call, what questions they asked, what objections they pulled up, what did they like, what pains did they, all the things that we normally do, but we want to take it from those real live calls. And then once we have that, the third agent is going to go out there and actually scrape for those signals. So that seed list that we found, that we used to find new clients, now I'm take that seed list, and I'm going to reduce it, filter it by buying signals.

Because as we know, only 3% of the market is ready to buy at any given time. So I want my sellers focused on higher probability deals. So let's say once a week, I do that automation for those scraping for those buying signals, those intent signals that I found in my call extraction, I'm go out and looking for those. So it could be new businesses open, it could new sales leadership.

I don't know what your signals are, but those are the ones that I'm scraping for and every Monday morning my sellers wake up, that call report is something that's in their Slack channel or email is in there for them to say, hey, of my 100 account lists, here's the ones that are actually dealing with something. So I'll pause there from a lead generation perspective. That sounds tremendous to me. So I am now very curious, like, what are we talking about here in terms of the tech stack?

I'll be honest with you, I am 100% using just Copilot for everything that I just described. Before I say this, I want to go back in because everybody tries to get into ninth or tenth grade way too quickly, into automation way too quick. And I have to say, guys, please, please always build individual first, all those agents that I went through. Make sure that they are working, they're providing the output individually before you move to what Brandon and I are about to talk about, and that's automation.

So these entire automation can be done in Copilot Studio. So there are workflows that you can build in there to say, to attach it to your CRM and extract data. And a lot of people will push back on this and is what if our data isn't clean? I don't care.

Start pulling what you can. Your goal is to learn. We're in a phase where we have to learn fast. And I'm like, no, identify three fields.

Like your name, address, it's like figure out what fields that you can start to identify and just start playing with it. And what you'll do is you'll start to realize, hey, this field in CRM has to be populated. We're gonna start to mandate that because I need to pull these opportunity records or I need to pull this specific object. You'll learn that.

So start small. But you can do all of this in Microsoft Studio, which are two different, I'm going to call them instances. They might be two different platforms that Microsoft does, but they can create those workflows. They have API connections, so you can every single.

Thing in there. So, half of our listeners are going to be Microsoft houses and be able to have access to that. The other half are going be Gmail, Slack, ChatGPT, and then maybe using Klay, we are looking at a new product called Get Cargo, which seems to be the more modern Klay and Klay is already so modern, it's like I feel sorry for Klay that it's getting displaced by Get Cargo. And then Zapier was already displaced by Clay.

But I'm guessing that they all have similar functionality to Microsoft Studio, so it's some Zapier automation. You're right. I am actually working a lot. So Zapier and then my favorite is Make.

There's N8n and then there's a few more that I've been playing around with, but I'm not quite ready to say their names because I'm still playing around with it, but I like it when I see. But there's lot of different, if you're not a Microsoft user, then yeah, go to N8 and go to Make. Any of these Zapiers that allows you to have one task connected to the next task works just fine. And here's what's beautiful.

When I say some of these tools, there are tools now that, and I wish I had enough experience with it that I can go, I highly recommend this one, but not yet, is there are schools now that they've written all the automation for you. You are literally dragging and dropping things like, you know, Gmail and do this and do that. And it is taking me like 10 minutes. And it does all of the Zapier make stuff for you, it's doing it all for you.

So it's coming quickly even if it's the one that you're not sure if you want to share you want to share go on preview us with it. It's relay.app and I think his name is Justin, the CEO. He has, first of all, an incredible story of how he has taken, let's just say, a large, let' say, 55 plus marketing organization and moved it down to one person with like 80 something apps doing the work.

So a really good story. And then he does a phenomenal job on YouTube teaching how to use his platform to do all these integrations. It's a game changer. Yeah, it's a games changer.

But again, I've only been playing with it for a week, so don't judge me. I think clay is going to be an incredible tool for a lot of organizations to get up and running. I started using it about a year ago, and I actually hired a consultant to do the work because I didn't want to try to figure out how to work clay the best way, but I'm not sure it's going to have legs for the long term. You know, it's my personal opinion.

Have a look at Get Cargo apparently. Obviously. Thank you for sharing. Because if we go back to your first two agents, you make it sound really easy.

So we have the seed list. Now we're going to go find ones that look like that. How does it find it? Do we need a database?

Do we a zoom info or an Apollo or whatever? Or does it just discover it on the internet? So this is like the honest truth is I had an SMB rep. So when you think about the SMB, it goes back to like Brandon, you're like, he got a scrape for it's like digging for a needle in a haystack.

So I had a SMB seller who had, I'm making this number, let's say 5,000 accounts and she would say 3% of them are good. Like, you know, I have a list of accounts, but she's like, I want more accounts like this one and this one, and this and we use this very and I know I love that you said you make it sound simple because I swear it is like it's so simple so I wrote this prompt and I think you have to really write really good prompts and I think you get good at writing prompts by just writing a prompt And so literally recording her to figure out exactly what that needle looked like in that haystack that she wanted and then taking those accounts that she had already closed that were good and then saying go find more of these.

We found and it took her a year of doing it. She said that she was like you know for the past year I haven't found any you know more accounts. We ran that. I had a 30-minute call with her and I will say in 20 minutes we were off the because it scraped and found her six new accounts to go after.

True, true, true. Every bit of that is just solid truth, is she now had six new accounts that were not in her CRM that she had never heard of that were ideal for her to go after. So I don't mean for it to sound simple, but I can do it. It really is just fun to go, ooh, what's a good prompt that will extract this type of data?

And you don't need Apollo or anything like that for this one that we just said. You literally just have to really know your ICP and get granular. And then you have to be super clear with instructions. And I love giving examples.

So those example accounts that she had had success in, giving those and saying, hey, the reason it was good is because they were opening new store locations, you know, they were expanding into new territory, like all those little details that we take for granted. Your agent. It's like food for them. Like, give it to me.

Don't be afraid to overshare with your agent." I love that I just said that. Don't be afraid. Overshare with your agent.

But it's crawling over the internet. And that's why I was asking, so it's just looking at companies in the internet that look like the seed list. Yeah, whatever your instructions say, you're exactly right. It is Curl in the Web.

And I will tell it things like, go to Yelp reviews. I'm being very specific about where I want it to crawl. You know, go the events page. Like, here, if I were telling you, Bethany, how to get to my house, and I just said, head north, like, help me out, Donna.

And so, I do, I tell a lot of people that agents are like really good recipes. If you want a quarter cup of sugar in there, you better say a quarter of a sugar level at all. You have got to get detailed if you want to have really good output. I'm a big pyramid user when I think about this.

So you guys mentioned your industry, your law firms, real estate, you've mentioned that. So that's that industry top layer. And then that account layer is what we just did with our seed list. So we went and we found lookalike accounts.

And then what we wanna do is we want an agent to extract the right personas for us. So each of the layers of the pyramid kind of get things a little bit more fine tune, fine tune fine tune. Now I'm gonna extract my specific personas that have that role and responsibility that is gonna align with the value that I'm looking at. And then I want to take that list, that lead list as maybe a CSV and upload it into the CRM and just call it my lead list and work from it before I convert it.

And then we've created an agent that knows good prospecting emails or what it is that the customers want to hear about. Yes, yes. And so I love how you're starting to do this is that, you know, I think the reason I found this pretty easy is I've always developed SOPs. And so for me to take those SOPs and move it into AI was like, oh, hallelujah, but I recognize not all organizations have taken the time to write these out.

And, so I would say best practice is grab an Excel sheet, Write those in there where you're saying industry, account, buyers, competition, and then looking at each one of those headliners and saying, okay, for personas, what do we need to do with personas? Well, maybe we need a scrape for the buying committee. We definitely need to outreach, okay? So we've got to outreach to different personas, different ways.

So, we want to create an outreach agent that knows our personas and we want specific emails that are written to those personas. If I'm going after somebody who has fiscal responsibility, then I want to make sure that I'm addressing that and I'm respecting their time and really knowing that my message is going to resonate with them. So, I think there's different layers that we have to look at for when we see our pyramid is to say, okay, what does really good look like? And just get started that way.

And then what's beautiful is that you go back and you're like, oh, okay. We just scraped our buying conversations, our really good accounts, and we just found that we totally missed a persona. So now let's take that persona, let's educate our agent on this new persona and what actually is resonating. And I know what's resonating because I went and scraped all my really great accounts that you're working with, right?

And when you talk about scraping to get the personas, because LinkedIn really blocks you, or does LinkedIn not block, how were you scraping? What are you scraping to these personas? I respect LinkedIn, but from a leadless perspective, you can absolutely download your entire first-degree connections as a CSV. Anyone who is connected to me, I can.

Scrape that or download it, my contacts from LinkedIn anytime I want. So that'd be the first thing I'd say. But then also, these people are out on the World Wide Web everywhere. They're doing podcasts.

They are in news articles. And so when I write my instructions for my agent, I am telling it, I want you to go like, do not stop at anything. I want to look at podcasts. I want you to look at news articles.

I want you to look at Yelp reviews. I want you to look at the about section of the website and I want you to find people who are talking about XYZ or people with this title or people who are you know discussing this challenge. And that's a scraping. The instructions go out and it's like a little agent that says, okay, let me go find what Donna told me to go find, I'll scrape that, wonder if she wants that, take that, and then it's gonna bring that list back and I can look at that list and go, I actually don't want that.

Rerun and here's some new, better instructions for you. And then when you're writing the content for the different personas, does each one end up being its own agent? So you have your finance content agent versus for me. I kind of keep things pretty simple.

So when I built my outreach agent, I have different variables that I'm asking. And I'm asked things like, what persona are we writing to? Do you know their pain point? And often that means just, hey, upload your discovery call or a discovery call, upload that.

It can extract information from uploaded calls as well. But in this case, it would be the variables of if you'll tell me who the persona is and what the pain point is, and I think there's another thing that's in there, then. I'll go do the work and it goes and does the work because of what's in its knowledge base in its knowledge base are personas and the persona cards you know that have all the psychographics of that persona and then the other thing that's in the knowledge base our email frameworks to say hey will you use um you know this framework for for them things like that so that i know that it's going to provide value to the CFO or COO or something like that.

So we have the agent that's going off. It's a ninth grade agent. It really knows the right personas, right accounts, right personas. We have our CSV file.

We're using them as leads. We have a great email writer, and now it's time to automate. Yes, I love that you let me get here. Is this when we get the 24-hour SDR?

Yes, and I won't give it the title of an SDR because I do genuinely believe in a 70-30 split, meaning I think that AI can do 70% of the work and that the SDR can do that 30%. And I really think that organizations need to move into that mindset of, in 2026, these department leaders will be managing both agents and people, like it's happening. And so, that 70-30 split there, so what non-revenue producing work is our agent taking care of, and then what is our next role taking care, whether or not that's SDR or AE.

And being able to say, okay, you are my automation layer. And what we're going to do is we're gonna take these three agents and we're gonna create a workflow that allows that to say go out in research, then go out and find the buying signals, and then go out and write an outreach, whatever that process is, the workflow, so whether or not you use Make or Zapier or Copilot Studio, it doesn't matter, but you're going to string those agents together. And now I say that I, as a human, initiate that trigger and let's just call it a call agent trigger, whatever you want to call it, a prospecting agent.

I initiate that, the workflow goes to each one of the agents, does what it's supposed to do, and I, as the human, then see what that output is and say, approved, go do. So, I think having that human element as a part of the workflow is critical to getting started. Human in the loop, particularly to make sure it looks good. Did I not say that?

Did I say it? He said, yeah, I'm just using the trendy term for it. I'm a big believer that all of the AI agent needs to tie to revenue. And so I think that somebody in rev ops should be the AI overseer to say, hey, what agents are we creating?

And I think an agent card needs to be created so you know what the input and outputs are going to be. And then I think there has to be that governance around those agents that get created. I just had another thought as an idea. So SDRs often were responsible for all the outbound, a lot of like being really smart on who you're gonna contact, doing a lot this manually in the past, and then also a lot cold calls.

I don't know in America, but in the UK, like cold calling still works. So there is a lot like dialing and calling. Do you think that in the future, the SDR role might get split between? The people who are doing all the building, automating and creating this air cover and making sure it works and people who just do coke holes all day.

Yeah, I think the role of an SDR is dramatically going to shift. And I'm seeing a lot of these SDRs asking questions around how do I improve my skill sets to be able to do what this AI is coming. So I think there's going to be a major shift. And I do think that we'll still have that cold calling that's there, but I don't even know if I would be guessing it.

I am horrible at predicting the future, but but I do see it evolving. Yeah, because I guess I just see from the SDRs I see in the world, the ones who are kind of loving the cold call, love the interaction, and the ones who are actually way more technical and get really into the process. Because it's hard to find somebody who wants to do a lot of process and hit the phones. They don't tend to be the same.

Yes, very true. So yeah, it'll be interesting. You know, there's AI agents that are now calling and they are really good. And like, that was fun.

I don't know. I don't. And that's why I'm here just to say everybody, like, I do know what's currently available. I do what we can currently take advantage of, like get started.

Like that's my mantra, get started. So I think you've preempted it, but you might just have to answer it again. Which is, if our listeners can only take one thing away from the episode today, what is it? Two word, get started.

Don't be afraid of it. Go in there and implement the three things that I said, your output, your instructions, and your knowledge base. And then see what it gives you. And if it's not what you wanted, go tweak the instructions and try again.

Get started, it's moving fast. On that note, I will get started on my target list that I'm going after for these law firms based on your guidance here. So thank you very much, Donna, for joining us on the operations room. If you like what you hear, please subscribe or leave us a comment and we will see you next week.