In this episode of In-Ear Insights, the Trust Insights podcast, Katie and Chris discuss the reality behind the latest AI buzzwords and why they are not actually new. You will discover how to use the 5P framework to structure your agentic AI prompts for better results. You will learn the importance of defining a clear checklist for your AI agents to ensure they complete tasks without errors. You will understand how to avoid common pitfalls like “vibe coding” by grounding your projects in established requirements gathering. You will gain clarity on how to stop chasing trends and start applying proven methodologies that drive real performance.
00:00 – Introduction
01:30 – The rise of requirements engineering as an AI buzzword
05:15 – Why the 5P framework solves the goal engineering problem
09:45 – Defining “Done” and using checklists for AI agents
14:20 – Risks of skipping governance in agentic AI
18:00 – Call to action
Watch this episode to master the essential requirements for successful AI implementation.
Watch the video here:
Can’t see anything? Watch it on YouTube here.
Listen to the audio here:
- Need help with your company’s data and analytics? Let us know!
- Join our free Slack group for marketers interested in analytics!
[podcastsponsor]
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 S. Penn [00:00]: In this week’s In Ear Insights, it’s buzzword bingo season. And Katie, I’ve got a new one for you that you’re gonna love. You actually will genuinely enjoy it because it’s 100 squarely in your wheelhouse. But it is just an extension of what we’ve been talking about and that phrase. We’ve gone from prompt engineering, which was 2024, 2023, 2024, to context engineering, which is 2025, which we’ve been calling knowledge blocks for quite some time. But now sort of in 2025 was more about context management. How do you work within the limits to the 2026 term because of agentic AI that the hype Bros have decided to call requirements engineering? But this is something that we have talked about again, actually, I think we’ve been talking about since 2020. Requirements Engineering is the 5P framework and it is about how do you tell agentic AI?
Christopher S. Penn [00:58]: And we covered this in the last two episodes of the podcast, but it’s worth, you know, revisiting. In requirements engineering, how do you tell agentic AI to behave and to work? And I saw this on LinkedIn and it kind of irritated me, people saying, oh, forget this. You have to learn requirements engineering now. And I’m like, but if you don’t know how to do prompt engineering and you don’t know how to do context engineering, you can’t also do requirements engineering. It’s like trying to start building the house on the third floor.
Katie Robbert [01:31]: That’s all I have to say to that is so. I mean, kudos to people who are like, finally figuring it out. But to claim that it’s like a brand new thing. Absolutely not. Like, the 5P framework is just a framework to help people get organized with requirements gathering. Requirements gathering is not new. It is centuries old. I was going to say centuries old. I was going to say tens of thousands. Yeah, it’s not a new idea. It’s essentially getting all your stuff together before you start doing the thing. Like, what do I need to be set up for success? That is not a new concept. For some reason, people keep trying to, like, reinvent it as, hey, I just had this brilliant idea.
Katie Robbert [02:19]: What if we gathered all of the ingredients before we start cooking instead of turning on the stove and then trying to figure out what goes into the pan and then everything’s burnt. Like, wow, light bulb. I’m so impressed with this new term, this new buzzword. Oh, my God. Like, can we not. Can we just, can we just do the work correctly and Stop trying to, like, call it something fancy, but you’re absolutely right. Like if you can’t do requirements gathering or Requirements Engineering whatever, then you can’t do prompt engineering because you don’t know what you’re putting into the prompt. Like, this is so silly. This is the. It’s not you personally, but this is the dumbest thing that you have brought forward in a while. Because requirements engineering, oh my God. Like, really, really.
Christopher S. Penn [03:18]: It is, it is all over LinkedIn. I saw people that we know in the AI space foaming at the mouth of it. I’m going, this is not new, guys. This is stuff that you should have been doing all along. However, I would say that our twist on is twofold. One, obviously the 5P framework by trust Insights, which we’ve been beating that drum for six years now about it because we. Katie invented it. Katie invented. Katie Robert, the CEO of Trust Insights, invented the 5P framework by trust Insights as a form of goal engineering. We’re going to come back to that in just a second. And it is the most powerful framework for getting AI and humans to achieve things.
Christopher S. Penn [04:07]: And the reason why this is coming up a lot now is because in all of the agentic tools, Hermes Agent, which we covered on the live stream a couple weeks ago, Claude Code now has this open. Code has it, I believe Codex has it, Anti Gravity has it. There’s this new single command called goal, and you say goal and then you give it a goal to achieve and programs the tool to go into much more of an autonomous mode and to stop asking for things, just try to achieve the goal. Hence goal engineering as a subset of requirements engineering. However, what people are figuring out is if you give it a vague goal, it goes off the rails really fast. So, like, whoa, we need requirements engineering for our goal engineering.
Katie Robbert [04:56]: Oh my. I just.
Christopher S. Penn [05:00]: Wow.
Katie Robbert [05:02]: You know what I’m. They’re getting, They’re getting a slow clap for that. Wow. I mean, here’s the good news. We could focus on the bad news, we could pick on people. But here’s the good news, number one, I feel validated because this is something that even before the 5P framework by trust Insights was a fully formed creation. Chris, how long have I been telling you need requirements before you start messing around with any development?
Christopher S. Penn [05:41]: Since literally the day you started on my team, 11 years now.
Katie Robbert [05:46]: 11 Before that, I managed development teams for about a decade and that was always the pain point with them is they would just start developing, be like, no, I know what I’m doing. I’m like, but you need requirements. How do you know when you’re done? How do you know what the end goal is? And prior to me doing any of this, again, we sort of said, people have been doing this for millennia. Like, I did not invent requirements. It is around forever. And so these people who are coming to the table saying, guys, requirements. But wait, goals, outcomes. That’s the key. That’s the thing. Okay, fine, great. Call it what you want, take credit for it if you want to, that’s fine.
Katie Robbert [06:38]: I’m just glad people are doing it and recognizing that it has to be done because it really gives us an opportunity to. To flex the muscle of the 5P framework by trust Insights. Because if we go back in time just a little bit, the 5P framework is built on the back of digital transformation. So I want to acknowledge that, and I know I’ve talked about it before, but the challenge with digital transformation is that it always led with the technology first. So people process technology. Technology was the thing that they were trying to transform into, like to include some sort of a new, you know, going from, you know, CD ROM to web version digital transformation. But then you have to sort of backtrack into, well, who’s involved? Who are the people?
Katie Robbert [07:28]: Do we have what we need to do this in a repeatable way? That’s a process. And the goal was digital transformation. That’s not a goal, that’s not an outcome. That’s what you’re doing. So. So the 5P framework is trying to take that idea that people know and that people are accustomed to and really bookend it with these strong pillars of purpose. Why are you doing the thing in the first place? Why are you digitally transforming from a CD ROM to a web version? Oh, because the technology is changing and that’s what our customers are asking for. And because it will make it more efficient for development teams to develop against a web version versus a CD ROM version, where the testing is a nightmare and costing us more than we’re making in profits.
Katie Robbert [08:18]: Really good reason why you go to the other end of the 5Ps with performance. That’s your goal. Did we save time, energy, and headache on testing with the web version versus the CD rom? Are we more profitable with the web version versus CD rom? Are our customers happy with a web version versus a CD rom? The goal is not digital transformation.
Christopher S. Penn [08:48]: And in the agentic AI era, so many folks are kind of winging the goals. And you see this when you see what people are typing in as a prompt for their goals on LinkedIn, they’re so bad, they’re so vague, they don’t have any true performance measures. So what does this look like? What is the. The what does it look like in an agentic world? It looks like this. And for those who are listening on instead of watching, I’ll read this off at the end of one of my work plans. This is for it. This is for the Hermes agent. I have a section called definition of done, which is the performance part of the five Ps and it says all of the following must be true for work plan completion.
Christopher S. Penn [09:33]: So not even telling it what the plan is for the overall thing, but just telling you, hey, you know, you’re done with this task when these 12 conditions are met and there’s a series of tests and says you have to have zero errors on these tests. You have to have, you know, 1,000 of this thing done. You have to have a CSV in proper format. You have to have a runbook. You have to have an architecture diagram. This is definition of done. This is in the work plan that we built because which is based on the 5Ps. And then my goal is when I use the goal is completed Workplan MD because all of the 5Ps are in the work plan itself. Instead of me having to say, what do I tell the agent? Tell us. No, you have to complete the work plan.
Christopher S. Penn [10:20]: And in the work plan is a definition of done, which means that it reads, it goes, am I done? Hey, here’s a checklist we talked about at the end of 2025, Alibaba Quinn, the AI company in China, said that the most powerful validation technique to get AI to do what you want is to have it use checklists. You give it a checklist and say, are you done? They read checklists really well. And so in our goal engineering and in our requirements engineering, if you provide checklists and say this is what the definition of done means, then your goal is one line, which is fulfill the work plan. As you know, that’s written out. And I had this running over the weekend on Minimax, the Chinese model. It did phenomenal. And this is not a powerful model. This is the cheap model.
Christopher S. Penn [11:15]: This is an inexpensive model. But because I use the 5P framework by trust Insights and I defined everything out and I said, this is your definition of done. It took a while. It said, am I done? No. Am I done? No. And then finally, am I done? Yes. I can check everything off the checklist.
Katie Robbert [11:37]: A couple of weeks ago we gave the example of, you know, go to the store and get some milk. And so it was a vague set of instructions that, you know, we, as humans, we can put together, like, what the next logical steps are. But machines can’t do that the same way because it could be to them a thousand different possibilities of what does that mean? Go to the store, get some milk, and then throw it in the face of the customer next to you, and then, you know, set the building on fire. That’s a possible outcome. It could be, go to the store, get some milk, and then just lay on the floor and contemplate your existence. That’s a possible outcome.
Katie Robbert [12:20]: We haven’t specified what the outcome is to purchase the milk, bring it home, and get it in the fridge, you know, and so it’s. It’s a silly example, but I feel like it’s really. We’ve really been able to use it to explain why requirements engineering and why goal engineering are important things. So think about it, you know, think about it. Something more tangible. You know, were put. My husband and I were putting together a ceiling fan over the weekend. It didn’t have a lot of parts. The parts were labeled. But we still needed instructions. We still needed to know, where are all the pieces? Do we have the right screws? Do we have the right connectors? There’s electrical involved. We needed to make sure that those requirements were in place in order to fulfill the goal.
Katie Robbert [13:12]: The goal being to have a ceiling fan installed safely with the wiring so that it doesn’t catch on fire when you turn it on. For me, that’s like, I like that little bit of extra risk management. So I like to be really specific with my goals. But then you have to have the requirements. So, like, do we have a ceiling? Does the ceiling have electrical running through it? Have we turned off the power before we start pulling wires out of the ceiling? Like, those are really important requirements that if you decide to skip the steps, you’re going to be in a world of hurt, you’re going to electrocute yourself, you’re going to have a whole lot of drywall all over your floor, maybe all over your bedding, who knows? And then you’re possibly going to be served with divorce papers.
Katie Robbert [14:02]: If you’re trying to do this with your spouse and neither of you are reading the instructions, These are important things to factor in. And I’m so. I’m making light of it, but in all seriousness, like, when I see the kinds of applications that people are creating, they pose serious risk when they don’t have proper data privacy. They don’t have proper security or governance. They’re just putting it out there. They’ve just Vibe coded their way to an application that asks people to input their PII or financial information or to buy something and there’s no protection. And I think that this is, you know, again, when I say let’s focus on the positive. If the Vibe coders are coming around to the fact that we need requirements and goals, great. I don’t care how you got there. I’m glad you got there in the first place.
Katie Robbert [14:58]: But if you want to get there faster, then I highly recommend checking out the 5P framework by Trust Insights, which you can find at TrustInsights AI 5P framework. It’s going to get you organized. And the other thing is, it’s not just for humans. The 5P framework is hands down, the best way to structure your prompts for agentic AI. As Chris mentioned at the end of last year, Quinn said, or the company behind Quinn said get a checklist. The 5Ps are a checklist and they are a 360 holistic way to look at everything you need to consider. So you have your purpose, what is your why? You have your people, who’s involved, who benefits. You have your process. How are we doing the things you have your platform, what tools do we need, including the machine that we’re using, and then your performance.
Katie Robbert [15:58]: Those are your outcomes, those are your goals. Those five P’s become your requirements engineering checklist and your goal engineering. Because if you fulfill every one of those tasks on your Checklist within the five Ps you’ve reached your goal. Period.
Christopher S. Penn [16:19]: Yep. And it even makes things like organizing your knowledge bases, which we’ve talked about. And actually we should probably do an episode about knowledge graphs on the live stream at some point because I am finding they are the best way to keep AI on the rails for like knowing what’s supposed to be doing. But you need to have all of that as part of your requirements in engineering and the 5P is the best way to do that because it keeps everything in one place.
Christopher S. Penn [16:53]: If you think about like enterprise AI, which is another topic that I want to discuss at some point, there’s it is such a vast topic and all of so much of what I’ve seen people doing with sort of enterprise AI, this and that is so woefully incomplete compared to the vastness of dealing with the enterprise that if you don’t have Things like the 5P framework by trust Insights, when you do requirements engineering and you do goal engineering you’re going to be missing stuff. So just simple stuff going down the list. Legal, regulatory requirements, right? Compliance, your risk management, data strategy and data governance, because that’s part of it. Network topology in a enterprise is wildly different because, like it’s a company, it’s a Corporation that has 82 offices, right? And it has 300,000 employees and has like 800 satellite offices.
Christopher S. Penn [17:57]: The very network topology of your company will dictate what AI is possible or not. Because you got to navigate that like, oh, we have an office in Kuala Lumpur, right? And it’s powered by Starlink, which is not necessarily a secure network. So how do you securely connect your AI to the office in Kuala Lumpur? That’s a big deal. That’s part of requirement. That has to be part of your requirements engineering. But if you don’t, you have the awareness to know that for your project is important, then you’re going to come up with a goal or requirements that are incomplete and probably going to get you in a whole bunch of trouble.
Katie Robbert [18:38]: It’s for, you know, to sort of debunk or demystify these terms. Requirements engineering, goal engineering. There’s going to be a fair number of people who are going to throw their hands up and say, great, I had to learn prompt engineering, now I have to learn these other two things. Nope, you sure do not. It’s stuff that you already know. It’s getting organized ahead of starting a project. Who do you have to talk to? What stakeholders? Do you have the data? Do you have access to the correct platforms? Do you know why you’re doing the thing in the first place? Ask why. Hey, what are we trying to do with this thing? Why are we looking at our Google Analytics 4 data again? What’s the question we’re trying to answer? I thought we already had icps. Why are we refreshing our icps?
Katie Robbert [19:29]: Those are basic things that the tech bros or whoever it is, I don’t care who it is, just knock it off. Are repackaging basic skills to make them sound more important, to make them sound more expensive, to make them sound like something that you should be paying a heck of a lot of money for. No. And I say that as someone who runs a company based on clients paying us to do things, don’t pay us to do, you know, requirements and goal engineering. That’s silly. We can certainly help you get organized. But if you were coming to us and saying, I need requirements engineering, I’m gonna gently, very carefully and respectfully tell you don’t need that. You don’t need that from us, you don’t need that from McKinsey, you don’t need that from anyone else.
Katie Robbert [20:25]: No shade to McKinsey, but, like, you don’t need that. It’s basic requirements gathering. If you are struggling with requirements gathering, we can absolutely help, but I am not going to teach you Requirements Engineering because that’s not a thing. It’s just not.
Christopher S. Penn [20:41]: LinkedIn says it is.
Katie Robbert [20:44]: Well, if it’s on the Internet, it must be true.
Christopher S. Penn [20:46]: Exactly. Well, so we do have the five levels of AI enablement. So I guess that means Trustee Insights should corner Role Engineering, which would be level four systems. Now we should probably corner, like, Charter engineering for Level 5 before anybody else does.
Katie Robbert [21:03]: Yeah, no, I. Why not? I mean, and here’s the thing. Like, you know, again, we’re sort of making light of it. We’re calling BS on where we need to call bs. But I also understand that there’s the people who are logical and then there’s the people who are caught up by the shiny objects. And we need to be able to cater to both. So if you need to go back to your, you know, leadership team or your board or whoever it is making the decision, and you need to call it Requirements Engineering or Goal Engineering, cool, Go ahead and do that. I’m happy to help you do that. If you want to call it Role Engineering, if you want. Whatever you want to call it. Honestly, I don’t care. But for your own sanity, know that you’re not learning something new.
Christopher S. Penn [21:49]: Yeah. And that there are already well established best practices for all these things. So you don’t have to reinvent the wheel.
Katie Robbert [21:57]: No, absolutely not. That’s just silly. Don’t do it. Use the 5P framework. You can go to TrustInsights AI 5P framework. It will give you. There’s, you know, free downloads, there’s, you know, walkthroughs, there’s examples, you know, create a checklist. Why am I doing this? Who’s asking me to do this? How do I do this? What platforms am I using to do this? And did I do the thing? You could also do the AI readiness assessment, which you can get at TrustInsights, AI readiness assessment, which will even sort of give you a baseline. The goal of this particular assessment, it’s a free assessment, is to give you an understanding of where you are in your AI journey in terms of your readiness or your company’s readiness to be using AI more readily, more heavily, more agentically.
Katie Robbert [22:57]: Like, you know, there’s a lot of things to consider, which you know Chris and I will be covering in the next few weeks. You know, there’s more governance to consider, there’s more data privacy to consider. There’s being able to get your data, there’s being able to get a consensus from, you know, the people in charge as to what you’re doing, what you’re building, who it’s for. But again, those are not new skills. Those are not new tasks that you need to do. AI is just amplifying the need for these things to be done correctly.
Christopher S. Penn [23:30]: Exactly. So if you’ve got some thoughts about requirements engineering or goal engineering and you want to share them, pop by our free Slacker Go to trust insights AI analytics for marketers where you and over 4700 of the marketers are asking and answering each other’s questions every single day. And wherever it is you watch or listen to the show, if there’s a channel you’d rather have it on instead, chances are we’re there. Go to Trust Insights AI TI Podcast. You can find us in all the places fine podcasts are served. Thank you for tuning in. I’ll talk to you on the next one.
Katie Robbert [24:06]: 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 Robert 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 Insight specializes in helping businesses leverage the power of data, artificial intelligence and machine learning to drive measurable marketing roi. Trust Insight 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.
Katie Robbert [24:59]: 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 metalama. 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 in 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 explaining complex concepts clearly through compelling narratives and visualizations.
Katie Robbert [26:04]: 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.
|
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! |
Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new episodes every Wednesday. |
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.
Newsletter
Podcast
