In-Ear Insights: How to Build an AI Strategy

In this episode of In-Ear Insights, the Trust Insights podcast, Katie and Chris discuss how to build an AI strategy. Discover why focusing on AI as a strategy is misguided and what you should do instead to achieve meaningful business outcomes. You’ll learn how to identify the actual problems you need to solve and how AI tools like generative AI can fit into your existing processes and strategies. Gain insights into the six broad use cases of generative AI and explore how to leverage these capabilities to enhance your business operations. Finally, understand how to effectively communicate and collaborate with stakeholders who may have unrealistic expectations about AI.

Watch the video here:

Can’t see anything? Watch it on YouTube here.

Listen to the audio here:

Download the MP3 audio here.


Machine-Generated Transcript

What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.

Christopher Penn 0:00

In this week’s In-Ear Insights, one of the things that Katie, you and I have talked about—and you have been correctly adamant about—is that AI is a tool.

It is a set of technologies.

It is not a strategy.

It is not something you lead with.

We lead with the five P’s: purpose, people, plan, process, platform.

First, what is the thing you’re trying to do? And is AI a good fit for it? But we do have to acknowledge the reality out in the jungle that is the modern business, which is that AI is the shiny object, and everyone and their cousins are like, “What is our AI strategy? How do we build our AI strategy?” You know, marketing director, whatever, “We need an AI strategy, go get me an AI strategy.” And we’re all like, “That’s, that’s really a bad idea.” But that’s reality.

So with that in mind—this is inspired by an actual customer inquiry we got this morning—when someone says, “I have to have an AI strategy, even if there is absolutely no purpose for it, but I need to make the corner office or the board happy.

I need to come up with a strategy.” Katie, what do we, what do we do to help that person deliver something that is an AI strategy, even if it is not relevant to the situation?

Katie Robbert 1:15

So you light some candles, you lay out some, you know, your herbs, and you start to chant and say, “Please make this go away.” And you hope for a celestial event or a full moon.

In reality, you know—and you’re right, you’re absolutely right—just because we know (you and I and, you know, a handful of other people) that AI is not a strategy does not mean that that’s not what’s happening in the market.

I used to have this, I used to have this boss, and he used to say this thing that I absolutely hated because it, to him, it sort of became like a mantra and excuse.

But in this particular instance, the phrase is correct, where he used to say, “Perception is reality.” And I hate that phrase.

But in this instance, that is what’s happening, because the perception is that AI is a strategy, not a tool.

Reality is that it is just part of your tech stack, and you need to figure out what it’s like.

Generative AI is akin to having a CRM or a marketing automation system or, you know, some sort of web analytics platform.

You can say, “What is our Google Analytics strategy?” You can say, “What is our Hubspot strategy? What is our CRM strategy? What is our email marketing strategy?” But at the end of the day, those are just tools, part of the overall plan.

Your strategy should be, “How are we bringing more customers into our ecosystem? How are we generating more awareness? How are we providing value?” And then we have our list of tools.

So all of that being said, Chris, I understand that people are saying, “What is my AI strategy?” so we can help them.

But we’re not going to put together an AI strategy.

What we’re going to do behind the scenes for them to bring to the C-suite or whoever’s asking for this thing is to walk them through the five P’s.

To your point, it’s purpose, people, process, platform, performance.

We already know the platform has been chosen.

We understand that it’s going to be generative AI, so we need to work around, you know, that decision to say, “Okay, so you’ve already chosen generative AI.

Let’s talk through what this is.” So what we’re doing is we’re still putting together a strategy with generative AI being the chosen platform.

And so it’s all just a matter of how you phrase it and position it.

If they want to say, “Our strategy is our CRM, our strategy is our marketing automation system, our strategy is generative AI,” fine, call it that.

But when you actually start to look at the pieces, the strategy is, “I want to generate more business with generative AI, I want to become a thought leader of generative AI.” And so you really, what you’re doing is you’re building a plan about being a thought leader or generating more business with generative AI being the tool.

I don’t really care what you call it, at the end of the day.

And that’s the thing—it doesn’t matter what you call it, the steps are still the steps.

Christopher Penn 4:31


And probably the easiest way for someone to get started—purpose is going to be pretty clear, right? Most businesses have the same general three purposes, right? You have to make money, save money, or save time.

Like, those are the three things that pretty much everybody wants.

I can’t think of a business where you don’t want those things.

So at a big picture, whatever the purpose is, it has to have a line of sight to at least one of those three objectives.

I think that’s probably fair to say.

So where you can start to make inroads in building an AI-centric magic is the process, looking at your existing processes right now and saying, “Well, of all the things that we do, what is, what are good candidates for generative AI?” And in many of our talks and things that we do, we have these sort of six broad use case categories of generative AI: making stuff with generation, extracting information out of existing data and making it more useful, summarizing, making big data into little data, rewriting, taking data from one format and turning it into another or translating it, classifying data, like, “Hey, we’ve got a bunch of stuff, we just need to get it organized,” and then question answering, “Hey, we’ve got this data, I want to get information out of it and explain it better.” So if you go back to your five P’s and say, “Okay, well, what processes do we have as marketers, as companies, as retailers, as salespeople?” And then you think about these different use cases, what processes most easily align to one of these six use cases?

Katie Robbert 6:05

It reminds me of an instance, or a situation, a scenario that we just went through with one of our clients.

A couple of months back, we met with their executive team, and they, because they had questions about generative AI, and they said, “What’s our, what’s our AI strategy for the year?” And we were like, “It’s not, it’s not, it’s one hundred percent not where you need to be right now because you have a lot of other things happening.” And we had to look, you and I had to look at a lot of different things before we could come up with some really good use cases for them, of where generative AI was appropriate.

And, you know, so like a lot of companies, they defaulted to, “Well, how can we create more content? How can we get rid of our writers? How can we generate more content with generative AI?” And for them, that was the wrong place for them to be.

So if we were to build a strategy around content generation, content marketing, it’s the wrong strategy because we’re looking at their processes.

And yes, there are a lot of inefficiencies in their processes.

But that wasn’t a problem that they needed to solve.

And so we could look at their process and say, “Sure, generative AI can help with this writing process, but that’s not your purpose.

Your purpose is like way over here in left field, you know, and you’re like at the soccer match, and so you’re playing baseball over here, but you’re trying to kick the ball through the net over here, and it just wasn’t gonna match up.” And so we had to look outside of what they were thinking.

And I think that that’s part of the challenge that a lot of companies are facing right now—that lack of understanding of what generative AI can do for them beyond that first use case, which is generation, because we get so stuck in, “Well, it can write for me.” It’s going to write stuff.

I was talking with… so I’ve mentioned before that I do some volunteer work at an animal shelter, and I was talking with one of the staff members, and she goes, “Oh, yeah, generative AI, like, it can like write for you, right? It can like write a blog post and like…” and it was a very simplistic understanding of it.

But for the general public, that’s their understanding—it’s going to write for you.

And so when someone comes to me and says, “What’s our AI strategy?” I know, immediately, they’re thinking, “How is it gonna write for me?”

Christopher Penn 8:38


We were talking before this episode about our SEO stuff as, as a company.

And one of the things that we’re going to do actually, on this week’s live stream—which, no matter when you listen to the show, you can always find our episodes from our livestream at—one of the things that these tools are really, really, probably the best at is actually summarization: “Hey, here’s a bunch of stuff, summarize it, boil it down.” And one of the things we’re going to do on the live stream is to take a look at all the data we have about SEO or technical site audit, our keyword list, what our existing content is, and actually have a tool summarize it and say, “What is, from your perspective as an outsider, Mister AI tool, what is the Trust Insights SEO strategy?” It may come back and say, “You suck at this,” which is exactly the kind of feedback that we would want to hear… no, not want, but we need to hear that, like, “Yeah, your, your SEO strategy is incoherent,” and then say, “Okay, well, well, then what should we do about it?” and we pivot to question answering, “What should we do about it?” at which point it might say, “Hey, do you have an example of someone who does it well?” Take that data, summarize it, and then say—do… it’s now moving to question answering—”say, well, here’s some companies doing it well.

Here’s you.

Here are the gaps in what you’re doing from a process perspective, based on what it can see if it, if we prompt it to say, ‘SEO is good, technicals is good, on-site is good, content is good, link building,” and then we hold up a good example and a bad example.

Hopefully, we’re not the bad example.

The tool can say, “Here’s, here’s what the differences are,”—summarization and classification—and it’s, in this case, also extraction.

So to your point, it has nothing to do with generation right now.

It is all about analysis, using these tools as really good analysts.

And that’s part of, part of the, I think, to your point, people think just all just generation all the time.

These tools have a lot more capability, and there are many, many more use cases than people are aware of what you can do with these tools.

Katie Robbert 10:48

I was thinking about this over the weekend, and you know, it strikes me because this is a really interesting conversation where you’re just listing off, “and then you can do this, and then you can do this,” when we started with the question of “how do I even use the thing?” I’m, you know, it’s sort of this interesting situation where there’s so many things you can do with generative AI to figure out what you should do with generative AI.

Like, it’s this weird, like, like nested, you know, enigma wrapped in a burrito kind of thing.

But you have to know the basics.

And, you know, it’s something I was thinking about over the weekend that I was gonna write about it for the newsletter this week—sort of the expectations of how to be able to, like, how fast you should be able to learn how to use something.

You know, and so Chris, you’ve been using and learning generative AI since its infancy, since before it became something that the public was aware of.

And so as you’re rattling off, “you could use summarization and extraction and then question answering and then pull this data in,” you’re making it sound very easy and straightforward when really it’s not.

You’ve had a lot of practice doing this.

And so that’s where, when someone comes to us and says, “Well, how do I fit AI into my strategy? What is my AI strategy?” Yes, that’s what we’re doing in the background, but we don’t expect them to be able to do all of those steps to figure out what their AI strategy is because we’re just using this as a tool.

It’s a tool.

It’s part of our toolkit.

I see that you have a reaction.

Please do.

Christopher Penn 12:35

I want to ask you this as a people manager.

Let’s say you have a new employee, the employee’s name is Gemma.

Gemma is your new intern.

Gemma is on her first day.

Gemma has a PhD in everything, like literally everything, a PhD in everything.

What do you do as a manager? How do you leverage Gemma’s incredible talents as a people manager? What would you first do?

Katie Robbert 13:00

Well, hopefully, I would first have an outlined problem statement that I’m looking to have Gemma solve.

You know, and so I would probably for myself start with my own five P’s.

And so, or at least some user stories and say, “Alright, I have this really smart person on my team now, what problems am I trying to solve?” First, so I would need to have my set of purpose statements.

And so I would say, “Alright, Gemma, I need you to do X, I need to find out the following information.

I just, I need to have a very clear direction in mind because clearly this person, you know, could go in a million different directions.

They could say, ‘Well, you know, I can tell you all of the state capitals, in alphabetical order, reversed.

I can tell you all of the state capitals by, you know, population size,’ like ridiculous things that this person could tell me.” And I could go down a rabbit hole of trying to see like, “Well, that’s really cool.

What about this? What about this?” but I’m, I’m not getting back to the purpose of why I engaged the person in the first place.

So I would have to start with my purpose, and then I would need to ask them, like, “What do you know about this? Can you help me solve this problem? This is a problem I’m having.” And they might say, “I don’t have enough information,” or “Yes, I absolutely can solve that problem.

Here’s how I would go about solving that problem.” So, you know, I’m understanding, Chris, that you’re asking me in the context of “how do I use generative AI,” but it’s very similar to how you manage people.

If you bring someone onto your team, you’re bringing them onto your team for a reason, because people cost money.

And so you can’t just hire people, pay them, and not have a reason… oh sorry.

You absolutely can do that.

You can do whatever you want.

I personally would never, have a reason, would never say, “You know what? I’m just gonna give this person my money.

I don’t know why I’m just gonna…” Okay, with that, Chris, we’re just going to start giving them part of our profits.

Christopher Penn 15:10

I mean, it depends on how many profits.

If we have so many, so much excess profit and we don’t know what to do with it, sure.

But until that day comes… no.

Katie Robbert 15:18

And even then we would probably still have to have some kind of a plan, like, “And I’m giving them this money because…” you know, and so it’s, working with generative AI is very similar to working with a new team member—you’re bringing them on for a reason.

So what is that reason? What is it that you need that new team member to do? What is their, you know, what are their KPIs? What are their performance metrics? How do you know this new team member is being successful? Who does this new team member need to interact with? Are they going to be an independent contributor? Or are they going to be part of a larger team where you’re saying they are one piece? They’re like thirty percent, and the rest of the team makes up the other information? How are they going to interact together? Is it something like, “Every morning at nine AM, we’re always going to check in and do the thing.

They are always going to bring the coffee and the doughnuts, they are always going to bring, you know, the data from yesterday so that we can review it together?” How are we going to get there? What are the other systems that we need to be thinking about? And how do we know we… how do we know that this new team member is fitting in and doing the thing and being successful and moving us forward? And those are all questions you need to be thinking about if you’re looking at bringing generative AI into your tech stack because it is just another team member, it is just another part of your tech stack.

You would have the exact same questions if you were evaluating a new CRM system.

“What is the purpose of having a CRM? What are we willing to pay for it? How much can we pay for it? What data do we have in it? What data can we get out of it? What reporting can we do? What is the process for keeping it up to date? Is it really cumbersome and it never updates? Can I attach my emails to it? Can I do this? Can I do that?” Those are all the same questions that you have about generative AI and your generative AI strategy—I’m putting this in air quotes if you’re listening.

Christopher Penn 17:16

No, it’s—That’s right.

That’s exactly right.

Because if you think about it, that’s the best way to phrase it.

How did you build your email marketing strategy? How did you build your analytics strategy? How did you build your SEO strategy? And if people think back to those questions and those deliberations, then they can realize, “Oh, it is around a technology that people have to use, and there has to be a process for using it, there has to be an outcome, or after, there has to be a reason for doing it.

If we do that, hopefully, it will be successful.” One of the things that struck me as you were talking about this case is that you were actually asking the exact questions that are in the Trust Insights PARA framework, which is at

It’s a free PDF, no form required.

But it really is, “What do you know about this topic? What questions do you have for me? What did I forget to ask? Did you do as you were told?” you know, for a new intern with two hundred fifty-five PhDs.

And I think it’s equally relevant that when we’re thinking about this crazy smart intern, would you have an intern with two hundred fifty-five PhDs going and getting coffee? Like, that’s probably not the best use of that person.

Would you have that intern writing blog posts? I mean, sure, you can, but you’re, you’re probably spending an awful lot of money on that intern with two hundred fifty-five PhDs just to have write blog posts, right?

Katie Robbert 18:39

And that’s exactly why… so to go back to the original question, when someone says, “How do I build an AI strategy?” the first question is, “What is the problem you’re trying to solve?” And if they say, “I need to have AI in my business,” that’s not a problem.

That is not a problem you’re trying to solve.

That is just a weird, shiny object, you know, flavor of the month.

That’s when, if this is a question that you are being asked, then it’s appropriate for you to go back and say, “Is this tied to a business goal? Is having generative AI as part of our overall strategy going to help us bring in more revenue, bring in more customers, retain customers, like whatever the, whatever the thing is?” Because if you can get to at least that, then you can start to work your way through the other four P’s to say, “Okay, at least I know my purpose is revenue.

Everything I do is going to circle around bringing in more revenue.” So if that’s summarizing data faster so that we can serve as more customers, that is a revenue-generating thing.

If it is rewriting our clients, you know, mission statements based on their information to be more or to their tone, that is something that would bring in more revenue.

But you have to know your why.

And that can help you build that, you know, that can help you figure out where generative AI fits into the rest of it.

But having a generative AI strategy just for the sake of having it is not a purpose statement.

Christopher Penn 20:19


And the easiest way to get that information out of a stakeholder is through a basic user story to say, like, “Hey, if you could just fill out this one madlib of a problem that you want someone else to solve with, with AI,” even if you lead with that because we know the stakeholder—their heart is set on having an AI strategy, we’re not going to change that—but if you can say, “Okay, here’s this template: ‘As a whoever you are stakeholder, I want to do this task so that you get this outcome.’ You can say to them, ‘Based on what you know about AI, which may or may not be a lot, could you fill out this madlib?’ And if they can do that, then you can start to pick apart and say, ‘Okay, well, here’s the one of the six use cases that this request fits in,’ and we can say, ‘Okay, yes, we can build an AI strategy around that particular use case and have it have an actual purpose,’ right? And if the stakeholder just cannot articulate a user story at all, then you can say, “Well, look, we can’t have an AI strategy just for the sake of AI because you couldn’t even fill out a madlib about it.”

Katie Robbert 21:28

Well, and, you know, I’ve run into this situation before, where, you know, you ask people what they want, they don’t know.

So that’s an opportunity for you to start putting together those user stories because people will be very clear about what they don’t want.

And so if you are running into, you know, so Chris, let’s say you come to me and say, “We need to be using generative AI,” and I’m like, “Okay, why?” You’re like, “Because we have to, because everybody else is.” That’s a cue to me to start putting together… so what you’re saying is, “As a company founder, I want to use generative AI so that I can stay competitive.” And so it’s an opportunity for me to start to try to understand what it is you’re getting at if you’re struggling to articulate it.

And you can be like, “Well, no, I don’t care about being competitive.” It’s like, “Okay, so as a founder, I want to use generative AI so that I can demonstrate thought leadership.” And you’d be like, “No, no, no, that’s not it.” So you can start to run through what you feel might be those business goals, until they’re like, “Oh, yeah, yeah, yeah, I want, we have to generate more revenue,” like, “Okay,” so there’s a way to get that information, even if the person you’re talking to isn’t being clear.

And a user story is a really great communication tool because it’s really simple.

It’s a, it’s a three-part sentence, and it’s the “so that,” the last part, that is going to give you the most information like for you, like, yes, you’re filling in the blanks on who they are, you already know they want to use generative AI, but it’s the “so that,” the purpose and the performance is going to tell you how the rest of this whole conversation is going to go.

Christopher Penn 23:16


Because if you don’t have those, those bookends, then it’s kind of like having a sandwich with no bread.

It’s like, “Okay, I’ve got lettuce, tomato, and bacon, but biologically, there’s nothing to hold it together.”

Katie Robbert 23:29

It’s a really sad Atkins sandwich, which is not sustainable, by the way.

Christopher Penn 23:33

That sounds terrible.

That’s exactly it.

So if you are in the situation where you are being asked to build an AI strategy or come up with an AI strategy, this is one approach to do it that will not only help you figure out, “Yes, here’s the six different use cases that generative AI fits into,” but also connect it to something that actually matters to the business so that when you look back in six months or a year, you can say, “Well, here’s the actual results we generated, and yes, we used generative AI to do it, but here’s the actual results we generated that we can, we can be happy that we, we engaged.” And it’s a way to take that perhaps misguided shiny object syndrome and use it to turn it into real results.

Katie Robbert 24:23

Yeah, I agree.

Christopher Penn 24:25

So have you been asked for your generative AI strategy? Have you had to come up with one? If you have, come over and share your experiences at—our free Slack group with over three thousand members asking and answering these questions every single day, including data, analytics, and, of course, AI—it’s been an all AI show everywhere for the last year now.

And wherever you watch or listen to this show, if there’s a channel I’d rather have it on instead, go to where you can find us on every major place that serves up a podcast.

And when you’re on the channel of your choice, please leave us a rating and a review—it does help to share the show.

Thanks for tuning in.

I will talk to you next time.

Need help with your marketing AI and analytics?

You might also enjoy:

Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly Trust Insights newsletter, INBOX INSIGHTS. Subscribe now for free; new issues every Wednesday!

Click here to subscribe now »

Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new episodes every Wednesday.

3 thoughts on “In-Ear Insights: How to Build an AI Strategy

Leave a Reply

Your email address will not be published. Required fields are marked *

Pin It on Pinterest

Share This