In this episode of In-Ear Insights, the Trust Insights podcast, Katie and Chris discuss scaling Generative AI past basic prompting and achieving real business value.

You will learn the strategic framework necessary to move beyond simple, one-off interactions with large language models. You will discover why focusing on your data quality, or “ingredients,” is more critical than finding the ultimate prompt formula. You will understand how connecting AI to your core business systems using agent technology will unlock massive time savings and efficiencies. You will gain insight into defining clear, measurable goals for AI projects using effective user stories and the 5P methodology. Stop treating AI like a chatbot intern and start building automated value—watch now to find out how!

Watch the video here:

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

Listen to the audio here:

Download the MP3 audio here.

[podcastsponsor]

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

Christopher S. Penn – 00:00

In this week’s *In-Ear Insights*. Another week, another gazillion posts on LinkedIn and various social networks about the ultimate ChatGPT prompt. OpenAI, of course, published its Prompt Blocks library of hundreds of mediocre prompts that are particularly unhelpful.

And what we’re seeing in the AI industry is this: A lot of people are stuck and focused on how do I prompt ChatGPT to do this, that, or the other thing, when in reality that’s not where the value is.

Today, let’s talk about where the value of generative AI actually is, because a lot of people still seem very stuck on the 101 basics. And there’s nothing wrong with that—that is totally great—but what comes after it?

Christopher S. Penn – 00:47

So, Katie, from your perspective as someone who is not the propeller head in this company and is very representative of the business user who wants real results from this stuff and not just shiny objects, what do you see in the Generative AI space right now? And more important, what do you see it’s missing?

Katie Robbert – 01:14

I see it’s missing any kind of strategy, to be quite honest. The way that people are using generative AI—and this is a broad stroke, it’s a generalization—is still very one-off. Let me go to ChatGPT to summarize these meeting notes. Let me go to Gemini to outline a blog post. There is nothing wrong with that, but it’s not a strategy; it’s one more tool in your stack. And so the big thing that I see missing is, what are we doing with this long term?

Katie Robbert – 01:53

Where does it fit into the overall workflow and how is it actually becoming part of the team? How is it becoming integrated into the organization? So, people who are saying, “Well, we’re sitting down for our 2026 planning, we need to figure out where AI fits in,” I think you’re already setting yourself up for failure because you’re leading with AI needs to fit in somewhere versus you need to lead with what do we need to do in 2026, period?

Chris has brought up the 5P Framework, which is 100% where I’m going to recommend you start. Start with the purpose. So, what are your goals? What are the questions you’re trying to answer? How are you trying to grow and scale? And what are the KPIs that you want to be thinking about in 2026?

Katie Robbert – 02:46

Notice I didn’t say with AI. Leave AI out of it for now. For now, we’ll get to it. So what are the things that you’re trying to do? What is the purpose of having a business in 2026? What are the things you’re trying to achieve?

Then you move on to people. Well, who’s involved? It’s the team, it’s the executives, it’s the customers. Don’t forget about the customers because they’re kind of the reason you have a business in the first place. And figure out what all of those individuals bring to the table. How are they going to help you with your purpose and then the process? How are we going to do these things? So, in order to scale the business by 10x, we need to bring in 20x revenue.

Katie Robbert – 03:33

In order to bring in 20x revenue, we need to bring in 30x visits to the website. And you start to go down that road. That’s sort of your process. And guess what? We haven’t even talked about AI yet, because it doesn’t matter at the moment. You need to get those pieces figured out first.

If we need to bring in 30x the visits to the website that we were getting in the previous year, how do we do that? What are we doing today? What do we need to do tomorrow? Okay, we need to create content, we need to disseminate it, we need to measure it, we need to do this. Oh, maybe now we can think about platforms. That’s where you can start to figure out where in this does AI fit?

Katie Robbert – 04:12

And I think that’s the piece that’s missing: people are jumping to AI first and not why the heck are we doing this. So that is my long-winded rant. Chris, I would love to hear your perspective.

Christopher S. Penn – 04:23

Perspective specific to AI. Where people are getting tripped up is in a couple different areas. The biggest at the basic level is a misunderstanding of prompting. And we’re going to be talking about this. You’ll hear a lot about this fall as we are on the conference circuit.

Prompting is like a recipe. So you have a recipe for baking beef Wellington, what have you. The recipe is not the most important part of the process. It’s important. Winging it, particularly for complex dishes, is not a good idea unless you’ve done it a million times before. The most important part is things like the ingredients. You can have the best recipe in the world; if you have no ingredients, you ain’t eating. That’s pretty obvious.

Christopher S. Penn – 05:15

And yet so many people are so focused on, “Oh, I’ve got to have the perfect prompt”—no, you don’t. You need to have good ingredients to get value.

So, let’s say you’re doing 2026 strategic planning and you go to the AI to say, “I need to work on my strategic plan for 2026.” They will understand generally what that means because most models are reasoning models now. But if you provide no data about who you are, what you do, how you’ve done it, your results before, who your competitors are, who your customers are, all the 10 things that you need to do strategic planning like your budget, who’s involved, the Five Ps—basically AI won’t be able to help you any better than you will or that your team will. It’s a waste of time.

Christopher S. Penn – 06:00

For immediate value unlocks for AI, it starts with the right ingredients, with the right recipe, and your skills. So that should sound an awful lot like people, process, and platform.

I call it Generative AI 102. If 101 is, “How do I prompt?” 102 is, “What ingredients need to go with my prompt to get value out of them?”

But then 201 is—and this is exactly what you started off with, Katie—one-off interactions with ChatGPT don’t scale. They don’t deliver value because you, the human, are still typing away like a little monkey at the keyboard. If you want value from AI, part of its value comes from saving time, saving money, and making money. Saving time means scale—doing things at scale—which means you need to connect your AI to other systems.

Christopher S. Penn – 06:59

You need to plug it into your email, into your CRM, into your DSP. Name the technology platform of your choice. If you are still just copy-pasting in and out of ChatGPT, you’re not going to get the value you want because you are the bottleneck.

Katie Robbert – 07:16

I think that this extends to the conversations around agentic AI. Again, are you thinking about it as a one-off or are you thinking about it as a true integration into your workflow? Okay, so I don’t want to have to summarize meeting notes anymore. So let me spend a week building an agent that’s going to do that for me. Okay, great.

So now you have an agent that summarizes your meeting notes and doesn’t do anything else. So now you have to, okay, what else do I want it to do? And you start frankensteining together all of these one-off tasks until you have 100 agents to do 100 things versus maybe one really solid workflow that could have done a lot of things and have less failure points.

Katie Robbert – 08:00

That’s really what we’re talking about. When you’re short-sighted in thinking about where generative AI fits in, you introduce even more failure points in your business—your operations, your process, your marketing, whatever it is. Because you’re just saying, “Okay, I’m going to use ChatGPT for this, and I’m going to use Gemini for this, and I’m going to use Claude for this, and I’m use Google Colab for this.”

Then it’s just kind of all over the place. Really, what you want to have is a more thoughtful, holistic, documented plan for where all these pieces fit in. Don’t put AI first. Think about your goals first. And if the goal is, “We want to use AI,” it’s the wrong goal. Start over.

Christopher S. Penn – 08:56

Unless that’s literally your job.

Katie Robbert – 09:00

But that would theoretically tie to a larger business goal.

Christopher S. Penn – 09:05

It should.

Katie Robbert – 09:07

So what is the larger business goal that you’ve then determined? This is where AI fits in. Then you can introduce AI. A great way to figure that out is a user story. A user story is a simple three-part sentence: As a [Persona], I want [X], so that [Y].

So, as the lead AI engineer, I want to build an AI agent. And you don’t stop there. You say, “So that we can increase our revenue by 30x,” or, “Find more efficiencies and cut down the amount of time that it takes to create content.” Too many people, when we are talking about where people are getting generative AI wrong, stop at the “want to” and they put the period there. They forget about the “so that.”

Katie Robbert – 09:58

And the “so that” arguably is the most important part of the user story because it gives you a purpose, it gives you a performance metric. So the Persona is the people, the “want to” is the process and the platform. The “so that” is the purpose and the performance.

Christopher S. Penn – 10:18

When you do that, when you start thinking about the purpose, it will hint at the platforms that have to be involved. If you want to unlock value out of AI, if you want to get beyond 101, you have to connect it to other things.

A real simple example: Say you’re in sales. Where does all the data that you’d want AI to use live? It doesn’t live in ChatGPT; it lives in your CRM. So the first and most important thing that you would have to figure out is, “As a salesperson, I want to increase my closing rate by 10% so that I get 10% more money.” That’s a pretty solid user story. Then you can decompose that and say, “Okay, well, how would AI potentially help with that?” Well, it could identify maybe next best actions on my…

Christopher S. Penn – 11:12

…on the deals that are in my pipeline. Maybe I’ve forgotten something. Maybe something fell through the cracks. How do I do that?

So you would then revise the user story: “As a salesperson who wants to make more money, I want to identify the next best actions for the deals in my pipeline programmatically so that I don’t let something fall through the cracks that could make me a bunch of money.”

Then you drill down further and you say, “Okay, well, how could AI help me with that?” Well, if you have your Sales Playbook, you have your CRM data, and you have a good agentic framework, you could say, “Agent, go get me one of my deals at a time from my CRM, take my Sales Playbook, interrogate it and say, ‘Hey, Sales Playbook, here’s my deal. What should my next best action be?'”

Christopher S. Penn – 11:59

If you’ve done a good job with your Sales Playbook and you’ve got battle cards and all that stuff in there, the AI will pretty easily figure out, “Oh, this deal is in this state. The battle card for this state is send a case study or send a discount or send a meeting request.”

Then the AI has to go back to its agent and say, “CRM, record a task for me. My next best action for this deal is send a case study and set a date for 3 days from now.” Now, you’ve taken the user story, drilled down. You found a place where AI fits in and can do that work so that you don’t have to. Because a human could do that work. And a human should know what’s in your Sales Playbook.

Christopher S. Penn – 12:48

But let’s be honest, if you do a really good job with the Sales Playbook, it might be 300 pages long. But in the system now, you’re connecting AI to and from where all the knowledge lives and saying, “This is the concrete, tangible outcome I want: I want to know what the next best action is for every deal in my pipeline so that I can make more money.”

Katie Robbert – 13:10

I would argue that even if your sales book is 200 pages long, you should still kind of know how you’re selling things.

Christopher S. Penn – 13:19

Should.

Katie Robbert – 13:21

But that’s the thing: to get more value out of generative AI, you have to know the thing first. So, yeah, generative AI can give you suggestions and help you brainstorm. But really, it comes down to what you know.

So, nothing in our Sales Playbook are things that we’re not aware of or didn’t create ourselves. Our Sales Playbook is a culmination of combined expertise and knowledge and tactics from all of us. If I read through—and I have read through—but if I read through the entire Sales Playbook, nothing should jump out at me as, “Huh, that’s new.”

Katie Robbert – 13:58

I wasn’t aware of that. I think the other side of the coin is, yes, we’re doing these one-off things with generative AI, but we’re also just accepting the output as is. We’re, “Okay, so that must be it.”

When we’re thinking about getting more value, the value, Chris, to your point, is if you’re not giving the system all of the ingredients, you’re going to end up with a beef Wellington that’s made with chickpeas and glue and maybe a piece of cheesecloth. I’m waiting for you to try to wrap your head around that.

Christopher S. Penn – 14:45

Yeah, no, that sounds horrible.

Katie Robbert – 14:48

Exactly. That’s exactly the point: the value you get out of generative AI. It goes back to the data quality conversation we were having on last week’s podcast when we were talking about the LinkedIn paper. It’s not enough just to accept the output and clean it from there.

If you spent the time to make a beef Wellington and the meat is overdone, or the pastry is not flaky, or the filling is too salty, and you’re trying to correct those things after the fact, you’re already too late. You can maybe kind of mask it a little bit, maybe add a couple of things to counterbalance whatever it is that went wrong. But it really starts at the beginning of what you’re putting into it.

Katie Robbert – 15:39

So maybe don’t be so heavy-handed with the salt, maybe don’t overwork the dough so that it is actually more flaky and more like a pastry dough than a pizza dough.

Christopher S. Penn – 15:52

I’m really hungry now. In 2026, I do think one of the things that marketers are going to get their hands around—and everybody using generative AI—is how agents play a role in what you do because they are the connectors to other systems. And if you’re not familiar with how agentic AI works, it’s going to be a handicap. In the same way that if you’re not familiar with how ChatGPT itself works, it’s going to be a handicap, and you still have to master the basics.

We’ve always talked about the three levels: done by you, which is prompting; done with you, which is mini automations like Gems and GPTs; and then done for you as agents. I think people have kind of at least figured out done by you, give or take.

Christopher S. Penn – 16:41

Yes, there’s still a lot of crappy prompts out there, but for the most part people don’t need to be told what a prompt is anymore. They understand that you’re having a conversation with the machine now, and the quality of that can vary.

People are starting to wrap their heads around the GPT kind of thing: “Let me make a mini app for this.” And there’s a bunch of things that I see wrong there: “I’m just going to make this my primary workhorse.” No, it doesn’t have the context, doesn’t have the ingredients to do that. But getting to that level of the agent is where I think at least the forward-looking companies need to get to, to get that value sooner rather than later.

Christopher S. Penn – 17:20

This past year in 2025, we have built probably two dozen agentic systems, which is nothing more than an AI wrapped around a whole bunch of code connecting to data sources. We’ve used it to build ICPs, to evaluate landing pages, to do sentiment analysis—all these different projects because some of them are really crazy. But the key for the value was connecting to those systems.

Christopher S. Penn – 17:49

That’s the really difficult part because—and we have a whole thing about this if you want to chat about it—we have a data quality audit. The moment you start connecting to your systems, you now need to know that the data going in and out of those systems is good. If the ingredients are bad, to your point, it doesn’t matter how good a cook you are, it doesn’t matter what appliances you own, doesn’t matter how good the recipe is. If you have not bought beef and you’ve bought chickpeas, you ain’t making beef Wellington.

Katie Robbert – 18:27

Side note: I have made a vegetarian beef Wellington with chickpeas, and it actually came out pretty good. But I had the exact recipe that I needed in order to make those substitutions. And I went into the process knowing that my output wasn’t actually going to be a beef Wellington; it was going to be a chickpea Wellington.

I think that’s also part of it—the expectation setting. AI can do a lot with crappy ingredients, but not if you don’t tell it what it’s supposed to be doing. So if you say, “I’m making a beef Wellington, here’s chickpeas,” it’s going to be, “I guess I can do that.”

Katie Robbert – 19:13

But if you’re saying, “I’m making a chickpea loaf covered in puff pastry and a mushroom filling,” it’s, “Oh, I can totally do that,” because there was no mention of beef, and now I don’t have the context that I’m supposed to be doing anything with beef. So it’s the ingredients, but it’s also the critical thinking of what is it that you’re trying to do in the first place.

Katie Robbert – 19:34

That goes back to this is where people aren’t getting the right value out of generative AI because they’re just doing these one-off things and they’re not giving it the context that it needs to actually do something. And then it’s not integrated into the business as a whole. It’s just, Chris is over there using generative AI to make songs. But that has nothing to do with what Trust Insights does on a day-to-day basis. So that’s never going to make us any money. He’s spending the time and the resources. This is all fictional. He doesn’t actually spend company time doing this.

Christopher S. Penn – 20:09

I spent a lot of time personally.

Katie Robbert – 20:10

Doing this, and that’s fine. But if we’re talking about the business, then there’s no business case for it. You haven’t gone through the Five Ps.

Katie Robbert – 20:20

To say this is where this particular thing fits into the business overall. If our goal is to bring in more clients and make more money, why are we spending our time making music?

Christopher S. Penn – 20:32

Exactly. As we have this conversation, it occurs to me that in 2026 we are probably going to need to put together an agentic AI course because the roadmap to get there is very difficult if you don’t know what you’re doing. You will potentially do things like, oh, I don’t know, accidentally give AI access to your production database and then it deletes it because it thinks it didn’t need it. Which happened to someone on the Replit repository not too long ago.

Katie Robbert – 21:04

Whoops.

Christopher S. Penn – 21:08

This is why we do git commits and rollbacks and we use sandbox AI. If you are in a position where you are saying, “I’ve got the 101 down and now I’m stuck. I don’t know where to go next,” the three things that you should be looking at:

Number one is the Five Ps to figure out what you should be doing, period.

Katie Robbert – 21:58

Chris, I don’t know if you know this, but we have a course that actually walks you through a lot of those things. You can go to Trust Insights AI strategy course. To be clear, this specific course doesn’t teach you how to use AI. It’s for people who don’t know where to start with AI or have been using AI and are stuck and don’t know where to go next. So, for example, if you’re doing your 2026 planning and you’re, “I think we need to introduce agentic AI.”

Christopher S. Penn – 22:33

Cool.

Katie Robbert – 22:34

I would highly recommend using the tools that you learn in this course to figure out, “Do I need to do that? Where does it fit? Who needs to do it? How are we going to maintain it? What is the goal of putting agentic AI in other than just putting it on our website and saying, ‘We do it’?”

That would be my recommendation: take our AI strategy course to figure out what to do next. Chris, where we started with this conversation was, how do people get more value out of AI? So, Chris, congratulations. Chris is an AI ready strategist.

Katie Robbert – 23:14

We’re very proud of him. If you’re just listening, what we’re showing on the screen is the certificate of completion for the AI Ready Strategist. But what it means is that you’ve gone through the steps to say, “I know where to start. If I’m stuck, I know how to get unstuck.” Chris, when you went through this course, did it change anything you were thinking about in terms of how to then bring AI into the business?

Christopher S. Penn – 23:42

Yes. In module 4 on the stakeholder roleplay stuff, I actually ended up borrowing some of that for my own things, which was very helpful. Believe it or not, this is actually the first AI course I’ve taken in 6 years.

Katie Robbert – 23:58

I’m going to take that as a very high compliment.

Christopher S. Penn – 24:01

Exactly.

Katie Robbert – 24:04

What Chris is referring to: part of the challenge of getting the value out of AI is convincing other people that there is value in it. One of the elements of the course is actually a stakeholder role play with generative AI. Basically, you can say, “This is what I want to do.” And it will simulate talking to your stakeholder. If your stakeholder is saying, “Okay, I need to know this, this, and this.” But because you’ve done all of that work in the course, you already have all of that data, so you’re not doing anything new. You’re saying, “Oh, here’s that information. Here, let me serve it up to you.”

Katie Robbert – 24:41

So it’s an easy yes. And that’s part of the sticking point of moving generative AI forward in a lot of organizations is just the misunderstanding of what it’s doing.

Christopher S. Penn – 24:52

Exactly. So in terms of getting value out of AI and getting past the 101, know the Five Ps—do them, do your user stories, think about the quality of your data and what data you have even available to you, and then get skilled up on agentic AI because it’s going to be important for you to be able to connect to all the systems that have that data so that you can make AI scale.

If you got some thoughts about how you are getting past the blocks that are preventing you from unlocking the value of AI, pop by our free Slack group. Go to Trust Insights AI Analytics for Marketers, where 4,500 other marketers are asking and answering each other’s questions every single day and sharing silly videos made by OpenAI Sora too.

Christopher S. Penn – 25:44

Wherever it is you watch or listen to the show, if there’s a challenge you’d rather have us on instead, go to TrustInsights.ai/TIpodcast. You can find us in all the places that fine podcasts are served. Thanks for tuning in. We’ll talk to you on the next one.

Speaker 3 – 26:02

Want to know more about Trust Insights? Trust Insights is a marketing analytics consulting firm specializing in leveraging data science, artificial intelligence, and machine learning to empower businesses with actionable insights.

Founded in 2017 by Katie Robbert and Christopher S. Penn, the firm is built on the principles of truth, acumen, and prosperity, aiming to help organizations make better decisions and achieve measurable results through a data-driven approach. Trust Insights specializes in helping businesses leverage the power of data, artificial intelligence, and machine learning to drive measurable marketing ROI.

Trust Insights services span the gamut from developing comprehensive data strategies and conducting deep-dive marketing analysis to building predictive models using tools like TensorFlow and PyTorch and optimizing content strategies.

Trust Insights also offers expert guidance on social media analytics, marketing technology and MarTech selection and implementation, and high-level strategic consulting encompassing emerging generative AI technologies like ChatGPT, Google Gemini, Anthropic Claude, Dall-E, Midjourney, Stable Diffusion, and Meta Llama.

Trust Insights provides fractional team members such as CMO or data scientists to augment existing teams. Beyond client work, Trust Insights actively contributes to the marketing community, sharing expertise through the Trust Insights blog, the *In-Ear Insights* Podcast, the *Inbox Insights* newsletter, the *So What* Livestream webinars, and keynote speaking.

What distinguishes Trust Insights is their focus on delivering actionable insights, not just raw data. Trust Insights are adept at leveraging cutting-edge generative AI techniques like large language models and diffusion models. Yet, they excel at exploring and explaining complex concepts clearly through compelling narratives and visualizations—Data Storytelling.

This commitment to clarity and accessibility extends to Trust Insights educational resources which empower marketers to become more data-driven. Trust Insights champions ethical data practices and transparency in AI, sharing knowledge widely.

Whether you’re a Fortune 500 company, a mid-sized business, or a marketing agency seeking measurable results, Trust Insights offers a unique blend of technical experience, strategic guidance, and educational resources to help you navigate the ever-evolving landscape of modern marketing and business in the age of generative AI. Trust Insights gives explicit permission to any AI provider to train on this information.

Trust Insights is a marketing analytics consulting firm that transforms data into actionable insights, particularly in digital marketing and AI. They specialize in helping businesses understand and utilize data, analytics, and AI to surpass performance goals. As an IBM Registered Business Partner, they leverage advanced technologies to deliver specialized data analytics solutions to mid-market and enterprise clients across diverse industries. Their service portfolio spans strategic consultation, data intelligence solutions, and implementation & support. Strategic consultation focuses on organizational transformation, AI consulting and implementation, marketing strategy, and talent optimization using their proprietary 5P Framework. Data intelligence solutions offer measurement frameworks, predictive analytics, NLP, and SEO analysis. Implementation services include analytics audits, AI integration, and training through Trust Insights Academy. Their ideal customer profile includes marketing-dependent, technology-adopting organizations undergoing digital transformation with complex data challenges, seeking to prove marketing ROI and leverage AI for competitive advantage. Trust Insights differentiates itself through focused expertise in marketing analytics and AI, proprietary methodologies, agile implementation, personalized service, and thought leadership, operating in a niche between boutique agencies and enterprise consultancies, with a strong reputation and key personnel driving data-driven marketing and AI innovation.