Sam and Steve discuss the practical applications of AI in modern sales and marketing. Steve explains his process for using Gemini’s Deep Research feature to generate detailed company reports, condensed infographics, and audio summaries for effective account-based marketing.
Listen to this episode here, on Apple Podcasts, Spotify, or wherever you get your podcasts.
Sam shares his techniques for leveraging AI to create personalized connection requests on LinkedIn and to repurpose podcasts into engaging social media posts.
They also explore how AI is changing SEO, shifting the focus from clicks to brand mentions within AI-generated search results. The conversation covers how AI can transform existing content into new formats and significantly shorten the traditional sales discovery process.
Transcript
STEVE: Welcome to CRM Talk, the show that brings you the latest in CRM and CRM related news and information. This is Steve Chipman, along with my co-host Sam Biardo, and today we’re going to talk a little bit about a topic you may not have heard about much about recently, AI, and how it pertains to CRM, marketing, and sales. How you doing, Sam?
SAM: Doing well, Steve. And you were telling me just before the show about how you’re using AI to do account-based marketing.
STEVE: I guess that’s what it is. So it started with you saying that you’re using Gemini through search and trying to make sure that your Starfish product appears in Gemini search results. And I haven’t actually used Gemini just for straight search. I tend to use regular Google, AI overviews, and all the other stuff on the screen. And then now AI mode. But what I’ve been using Gemini for is doing account research for salespeople, sales teams. And I’ve got this fairly elaborate prompt, and I use Deep Research. So for those of you who are on Google Workspace, if you’re on Google Business or higher, this doesn’t work with Google Workspace Starter, but with Business and higher, you have this mode in Gemini called Deep Research. This is in Gemini 2.5 Pro.
And what you can do is feed it a prompt telling it who what your company is, what you do, and then asking to research a specific company and feeding it parameters such as I want to learn about executive management, financial information, go-to-market strategy, technology stack, if you’re in the CRM business like we are. And it comes back with a very elaborate report.
Now, what I do for Salesforce users is I’ll put that into the account object. If you have Salesforce connected to your Google Drive, then it’s pretty easy to create a custom object that lets you take this Gemini output, save it as a Google Doc, and then make it appear in the account record in Salesforce.
But Gemini Pro doesn’t stop there. What it also gives you is this canvas mode. And canvas is effectively just HTML output. And it’s a condensed version of this very long report. So, as you know, Sam, not all salespeople are going to read a 5,000-word report about their company, even if they were going to make a lot of money by selling into it. So with this canvas version, you can ask it for an info graphic or a condensed version of this long text document that’s nicely formatted. Now it’s in HTML and there are techniques for getting it into Salesforce, so it also is embedded in the account object. That’s probably the case with other CRMs.
And then the third thing I wanted to mention is Google developed under its Notebook LM group this concept of an audio overview, which you may have heard. We actually used it to wrap up a podcast a few episodes ago. And that what that does is it takes some content you feed it and it transforms it into a conversational podcast between a man and a woman. It’s a little bit hokey, but what I found with this, very interestingly, is that I was using this this method to research a lead we got, and it gave me the long document. I read it, my eyes were glazing over halfway through. I later listened to the to the hokey podcast rendition of the document, and I picked up on a key piece of information auditorially that I would not that I did not visually see. So, two things, two two reasons to use the audio version. One is some people are just naturally audio learners, so it’s better for them to learn that way. And two, even if you’re of the type that might read 5,000 words, you’re probably going to miss something reading that much information. And the audio version could could give you a key detail that you might use in your approach to that client when you’re selling. So those are just a couple of the ways that I’m I’m using Gemini. And Sam, I’ll let you talk a little bit more about
SAM: I I was just wondering if you could take the canvas document and send it to an AI and say, create us an action list of items I need to do to be able to sell into this account.
STEVE: You Well, you could do that as a follow-on prompt to the main document or or yeah, or off the canvas document. But you’d probably want to do it in the same in the same chat.
SAM: Yeah, see, we’re we’re we’re right now doing our account-based marketing through LinkedIn. And so the way we’re using LinkedIn is we’re doing the research about a company and all their executives and then we’re actually asking it to give us a 200 character or less, because that’s what LinkedIn requires, introduction to this contact so I so that I can connect to them.
And I’m using that to connect to people I really don’t know, but it really sounds personal because it’s pulling in data about the company and what this person’s role is and what their role is with the particular potential internal projects and it’s a little bit more of a personal connection. So we’re, you know, it’s not all it’s not always perfect. Not everyone connects to you because connections are crazy right now. Everyone’s connecting to everybody in LinkedIn. But the reality is is that that’s step one. And then what we do is we produce videos and podcasts. And what we actually have done is fed those into our an AI and then take the podcast and say, given this podcast, write a LinkedIn introduction, and then I’ll post the podcast out like we do CRM talk, but then I’ll have a LinkedIn post that links to that podcast that’s actually all been AI generated. And it gives a nice little quick summary of the podcast, and I don’t have to do that anymore.
STEVE: Wait, and who who are you sending that summary to?
SAM: I’m just posting it on LinkedIn. So, okay, yeah. So we’ve got the CRM talk, and once we once we post this, you’ll see if you go look on my LinkedIn page, you’ll see a message saying, “Go listen to our podcast. Here’s what we talked about.” But that’s not generated by me. That was generated by the AI listening to the podcast.
STEVE: Got it. Okay.
SAM: Yeah, and then it’s it’s great. I can’t tell you how many hours a day I feel like I’m saving by having what I would call the unimportant emails that you have to write being written by AI.
STEVE: Absolutely. And and I I said recently, I recently posted on X, you know, as everyone’s moaning about the decreasing clicks because when people get their answers through AI overviews or AI mode or Chat GPT, there are fewer people clicking through to the websites because, of course, all that information is has been consumed by the AIs and is and the answers are being provided in in many cases.
SAM: Right. But that’s why you have to train the AI. So if you understand that the AI is really reading what content there is on the web, you can you can force the AI under certain search criteria, such as what is the best way to integrate, one we were doing for Starfish was, what’s the best way to integrate Info CSI with Salesforce? And it would come up for a while there and say Starfish, our Starfish product, until it didn’t one day. I would check it like every couple of days and then it came up with Zapier. And then what we did is we wrote a bunch of articles and LinkedIn posts that amplified that. And then the next day it said, “Products like Starfish and Zapier.” So, so you can trick them, but you’ve got to be very specific and you need a very specific, say, blog article and a bunch of them that Google will then index and put in their summary. So the trick is, in generally, is to if you have a very specific question that people are Googling and you want that to come up in the AI, you need to have a series of supporting materials that are posted in different sources, such as LinkedIn, and and X and other, Facebook potentially, and other places that talk about this and blogs on your website and blogs that other people post and amplify. Do that and you’ll get and you’ll get into that search engine.
STEVE: But but what you you just said is more about getting your brand to appear. So, so what I’m hearing is the the clicks are down no matter what you do, but the key is to have your brand being mentioned. So you just, in your case, it was really a brand mention of Starfish. People aren’t necessarily immediately going to go to a specific article like in the 10 blue links days and find out more, but now they’ve learned about Starfish and they might go to starfish.com an hour later, the next day. So, so the the key that I’ve been hearing about AI overviews and Chat GPT and AI mode is it’s it’s really a brand-first SEO strategy or geo strategy.
SAM: Yeah, and and it’s it’s it’s sort of like SEO was, right? We we’d write all this SEO so that we’d be on that first page, right? And then and then now Google’s effectively replaced that, right? So when I do a search, you know, first the AI come, well, first the sponsored comes up, then the AI comes up, then that magic Google box comes up if it, where if you if it thinks you’re going to buy something, and that’s all being managed by Google, right? And and then the standard SEO comes up somewhere on page two. And who goes to page two? So, the the challenge is how do you get yourself into the AI answers. And that that’s essentially the same problem we’ve had before. It’s just a lot of SEO, but now it’s digestible SEO that the AI has to digest and put yourself into that answer set that’s being generated by the AI.
STEVE: But if you don’t have a strong brand, if you’re if you’re a startup and or if you’re a more established company without a unique name, you’re at a disadvantage in this in this new world in terms of being come coming up in in search results, AI search results.
SAM: Absolutely. Yeah, I totally agree. Yeah.
STEVE: So, so the other thing I was what I was going to say earlier on on X is that the post I said was that the the Google giveth and the Google taketh away. And so Google has been taking away clicks, but what they’ve been giving us are these incredible time-saving tools. And they’ve also been giving us these I I almost look at some of Gemini as as a content redistribution tool because if you have a blog post that you want to turn into a video, you can do that with a Notebook LM video overview. You could do it, if you have some production time, you could do it with a string of Veo 3 clips. So you can, so now Google’s giving you these tools to take all that content that you’ve developed over the years that fewer people are going to go to, and now you can you can repurpose it in slide decks, audio, video, you name it. And the distribution game is at least as important as it was before all of these these clicks were taken away. So that’s really important to keep in mind when you’re marketing your business.
SAM: Yeah, the other thing you can do is if you want, so I I always look at, I always think in PowerPoint. So, unfortunately, I’ll I’ll before I do anything, I’ll create a bunch of PowerPoint slides and then multiple versions, refinements, I have a nice little deck. But you can take that deck, once you feel you’ve got it in the right place, and you can send it to like Chat GPT and say, “Given this PowerPoint deck, write a document, create a document that describes what’s inside the deck.” And it will it will create a pretty accurate document of what you put in your in your sales deck. And so then at that point, obviously, you still have to read and edit it, but once you’ve done that, you can then post that up as an article on your website. And then, of course, that article gets indexed and reinforces the AIs that are producing those answers.
STEVE: You can almost look at AI now as an anything-to-anything translation tool when it comes to content. Any any format to any format translation tool.
SAM: Yeah, I I have, I’ve got a a a step-by-step guide and I literally for implementing marketing that we put together for a very specific product called Creatio. And I’ve sent that off to Chat GPT and said, “Create a PowerPoint presentation.” And it it it wasn’t great, but it wasn’t bad either. And so those are the types of things that I think AIs are going to excel at.
Right? So we we create some initial content and then we have it transform it into multiple formats. It’s still not really good at, I mean it’s good, but it’s not really good at saying create original content. It’s going to give you something that someone else did, and nothing’s going to be original in it. But if you have some original ideas and you can put that in one format, you can easily transform it, like you said, to a podcast, you can transform that to a PowerPoint or a Word document or a blog article or a summary on LinkedIn.
STEVE: Yeah. And and now, so if you’re a whether you’re a B2B or B2C business, it used to be for for smaller companies, it was absolutely out of the question to have a 45-second or 1-minute promo or commercial filmed and produced. Well, now you can do that with these video tools. I happen to think that that Vayo 3, Google’s Vayo 3 is the best. And what what it’s limited to now are 8-second clips, but you can get really creative within that 8-second constraint, and you can put together clips that are different scenes. You can also create character and scene continuity by grabbing the last frame of an 8-second clip and uploading that as a photo as part of your prompt when you give the prompt for the next scene. And I’ve I’ve done that to create character continuity in kind of a spoof I did on a Got Milk ad, I call it Got Got CRM. And I wanted to have the same character appear in clip two, six, and seven. So all I had to do was take the still of the character from an earlier clip and and Vayo reproduced it perfectly. Now, the the one thing about Vayo 3 is that I know on Google Workspace Business, which we’re on, you’re limited to three clips per day, which is pretty limiting because you have to do retakes just like in live video. And if you want to have unlimited or virtually unlimited clips per day, then you have to pay the big bucks for the full Google Studio. But if you’re on Google Workspace Business or higher, I encourage you to to check out Veo 3 videos. And a little bit of a workaround, I happen to have access to three different Google accounts. I got my own, I’ve got one for marketing, and one for admin. So I can actually do nine per day, but still not enough.
But if you’re of the mind and you want to promote your product or service and you’ve and you don’t mind spreading out the effort over several days, you can do it. The key is to build out your scenes first, build out a script, build out your scenes, and you can just start with a simple 8 to 10 scene promo, create the prompts for it, have a tool where you where you can grab the still of of one of the frames in the video. I happen to use Camtasia. And then with a little bit of elbow grease, you can pretty easily put together a 40-second or longer promo for your product or service that looks like something you would have paid $100,000 for 10, you know, 5 years ago.
SAM: Yeah, we’re experimenting right now with just using an iPhone. So I’ve got an iPhone on a tripod and I come up with a, we have come up with a question, someone in marketing came up with a question and some bullet points that they want me to mention. And I do like a minute and a half discussion on that one question and I’m done and we’ve been posting those. And I think those are becoming real popular now because it’s less than a minute. And I’m, I’ve, I’ve created a list of about 40 topics that I’m going to try to do all under 60 seconds, where I can, it’s like what’s the relationship between risk and length of project time, things like that. And and hopefully, and the reason why I’m they’re so micro is because again, if someone’s searching in, if someone is searching or asking that to an AI, the question is going to be, “Well, what, you know, what’s the relationship between X and Y?” And if you have that specific topic in a video, it’s going to be, the it’s going to be the one that they are going to find first and it’s going to also influence the AI. So again, that’s why I’m doing all these little mini topics because if someone asks that question, like, “How what’s,” they’re not broad questions, right? Like they’re not like, “What’s the best way to do user adoption?” Right? It It’s like, “How important is management in in user adoption?” Right? It’s got to be a very narrow band question, but those are the ones that people are asking the AIs. So, if you have the answer to that, you’re going to you’re going to be digested and come up as part of that answer.
STEVE: Well, someone pointed pointed out on X the other day that there are more YouTube videos in in AI, it was either AI mode or AI overviews, maybe both, than ever before. So that type of information probably should be on YouTube in addition to social.
SAM: Right. And, so I I just think this is like a great tool for creating your brand, your personal brand, your company brand, and conveying very specific type of information about your products. And then, as we talked about for account-based marketing, it’s a great way of of of finding your customers and going after them and connecting to them and providing content on LinkedIn and other sources that once you’re connected with them, they’re going to start getting impressions. And so that just again, proves that you’re an expert in their industry.
STEVE: Right. And then beyond that, when you start the sales conversations, if you’ve got that that AI-based deep research, you’ve got some insider baseball. So you you you don’t ask as many questions that are have obvious answers because you know already and you’ve got some insights into their organization. Let’s say you’re selling technology. Let’s say you’re selling CRM like we are. Well, I found this deep research, if you specifically ask about their tech stack and you you enumerate categories like ERP, it’ll actually come back and tell you what the ERP they have is. So a lot of times that’s a mystery, but they’ll and they’ll find that information through job postings, for example. So, think think about how long it would take you to figure out what ERP system a company had by telling a person to go through all the job postings for the last 5 months and see if there’s anything that that relates to hiring a finance person, for example. Now you get it right away.
SAM: Yeah, it it it’s interesting because, maybe even as five years ago, as of maybe five years ago, I had this massive set of questions, and I mean like massive, I mean like 20 pages of questions that would cover discovery. You know, and it would be like, “Do you,” you know, it would be 15 questions on knowledge base, you know, and 10 questions on quoting and tons of questions on different types of integrations that they would want to do, you know, integrations with ERPs, integrations with marketing automation, integrations with e-commerce, and so on.
And I’ve abandoned that. The questions, I still need to know those things, you know, and I was I was using it as a template for years when I was talking to a client to say, “Okay, they answered this one, they answered this one.” But I could literally give that question to an AI and a name of a company and get 90% of that back now.
STEVE: So AI has replaced most of discovery.
SAM: Correct. Well, it it it it augments, it changes discovery from, “What tech stack do you use?” to, “Hey, we think this is your tech stack. Is that correct?”
STEVE: Yes, exactly. It’s just confirmation now as opposed to starting from scratch.
So the other thing in terms of ABM I I tried out recently, again through Google Gemini 2.5 Pro Deep Research, was I said, “Investigate the management hierarchy of this account and return the results to me in this format.” And I uploaded a CSV, a model CSV, which happens to be the format that Lucidchart uses for importing. So, and it worked really well. I I know like anything else, just like you have to rewrite some of the text that comes out, you can import this exported CSV into Lucidchart, builds out the hierarchy for you. And if something’s not right, you can drag stuff around. Or, as you say, the confirmation, when you’re on the call, first call with the person, you can say, “Well, it looks like your your CMO is this person and these two people work for that. Is that correct?” And they’ll either say yes or they’ll correct you. But you’ve got this great starting point of a hierarchy. And even if you’re not a Lucidchart user, Lucidchart just happens to be a good tool, you can create a version of that in that in the canvas tool I mentioned earlier. So you can get the HTML output of the of the org chart. It’s a little funky. You have to give it some extra prompting to say, “You know, it’s okay to go wide, but don’t cut,” If you don’t tell it to put in scroll bars, it’ll it’ll cut it off at the side. So you sort of you learn as you go in terms of some of the vagaries of how these things output graphics, but you can get a pretty good rendition of an org chart.
SAM: That’s really cool stuff. You know, I have a pet peeve though. I have a pet peeve when when people don’t modify their AI generated stuff and you can recognize it immediately. So if someone sends me an email that says, first sentence is, “I hope this finds you well,” which is what Copilot says, I just don’t even read it. It just go, I just like, “Oh, delete.” So my my advice is, if it’s a really important message, don’t make it obvious that an AI read it because most people aren’t, most people are getting wise to the opening sentences from Chat GPT and Copilot and some of the others.
STEVE: Well, you can do a lot of negative prompting in your prompts and say, “Don’t open with things like this, this, and this.” And I see that more and more where, and you and you tell it more and more to “No fluff, sound like a human.” And it it actually listens to you to some extent if you give it those those prompts.
SAM: I wish it would stay there though, because what ends up happening is you give it to those prompts and it solves it for that, but I want that for all prompts, you know, going forward.
STEVE: Well, that’s that’s the next phase in these in these, they aren’t even calling them chatbots anymore, they’re calling them AI assistants. Right. So in these AI assistants, it’s the continuity session to session where it remembers the things you asked for in earlier sessions. I think Chat GPT is supposed to be doing that to some degree.
SAM: No, they are. They are doing that right now. The, I will say that, yeah, we’ve been playing with our Starfish company with interactive chat chatbots. And it’s it’s quite interesting what you can do, how quickly you can build out an integration if you just say, “Build an integration between this table and this table,” and then come back and say, “Oh, no, and change this to do this, and change that to do that.” You can knock out like a thing that used to take us a day, you know, could be under 2 hours now. Right. And it’s like it’s just going to get faster.
STEVE: Yeah. And it all comes back to spending the time. As humans, the best thing we can do is to become better prompters and to catalog all of our prompts. I don’t know about you, but I’ve got a couple, I have three different Google Docs. Each of them has about 30 tabs of prompts. And I used to just, you know, prompt prompt an AI assistant and then that would go in the ether. I’d forget about it. Now I make sure to document everything so I can always go back to it. And that’s
SAM: Actually, I’m going to steal that idea because I don’t do that at all. I’m I’m everyone’s a bespoke prompt. So I’m going to have to figure out how I want to manage that. But that’s a really great idea, Steve.
STEVE: Yeah. I I even thought there that there might be a database app for that, but that’s going a little too far. You can just, you can just do it for now. But over time, I think a lot of us are going to end up with hundreds or thousands of different prompt variants for different things.
SAM: I’m just I’m just thinking about it. Yeah, no, I could see I’ll need a bot to manage my bots.
STEVE: Exactly. I’m sure someone’s going to come out with a prompt management system. Yeah. Maybe Maybe I’ll ask. There’s this new app called Mocha, this app builder, which is very good, and it’s got built-in database. And I’m a I’m putting together, well, I had been putting together a pretty elaborate prompt to have it build out a CRM system. But then I realized, well, I’m just using the traditional CRM motif of accounts, contacts, leads, opportunities, tasks. And I realized that if anyone who’s building a new CRM system today, they’re making it native AI, and it’s not going to look anything like the old forms and lists that came out with when all these products came out a few years ago. So, got to got to think AI when you’re building, you have to think, anytime you’re building an app, you almost have to make it an AI-first app app. And then that that has its own set of complications, such as, “Okay, what are you going to, what are you going to connect to, and what’s it going to cost?” So.
All right. Well, with that, Sam, I think we can wrap up this episode, unless you have any other parting thoughts.
SAM: I’m good, Steve. Nice talking to you. We haven’t we haven’t talked in about a month. You had a wonderful vacation, I heard, and looking, yep.
STEVE: It’s good. Good talking to you, too, Sam. And to our audience who who are who are still hanging in there, thanks for listening. I know there’s been a little bit of a little bit of a gap since the last one, but, you know,
SAM: Conflicted schedules, is what I’ll call it.
STEVE: All right.
SAM: Take care, Steve.
STEVE: Take care.