Confident Isn't the Same as Correct: A Look at Governed AI-Assisted Selling

Our next guest is phenomenal. Tim Riesterer is the Chief Strategy Officer for Corporate Visions. So welcome, Tim.

Thanks, Gerhard. Always a pleasure to be with you.

So this is an obscure title. Confidence is not the same as correct. What did you mean by that?

I hope it's thought-provoking. Um, but we believe like AI-assisted selling, or some people are calling it AI coaching, um, is now a more ubiquitous thing. Um, but we believe it's at a crossroads. Uh, and that crossroads is it's giving super confident answers. AI is always very confident. Um, but the question becomes, is it giving the correct answers? And Only the wisest, most discerning, highly effective, judgment-oriented salespeople can tell the difference, and that's too often too little of the selling population. And and so a lot is happening where there's a lot of confident sounding cognitive surrender happening, and and it's not necessarily moving any needles. And so AI coaching's at a crossroads, and today I want to talk about how to. Get over that hump and get us to a place where it's really helpful.

Can you maybe start with an example?

I could start with an example, as if you knew my slides almost, Gerhard. So here's my visual. What I want you to focus on here is, is I've got Copilot. I'm circling. There's so many scribbles you can hardly tell. So here's Copilot. So we were working with a client that was using Copilot as their AI. It could be Claude, it could be Chat, pick your AI bot that your sellers are using. And so they put Copilot in a place where the sellers could ask it questions. And initially over here to the far right, they put a proprietary LLM in place that had product information, pricing information, all kinds of solution configuration content. But then they integrated their Seismic messaging and positioning and their Gong conversation intelligence and call transcripts. And their Salesforce for deal context. And this visual kind of says it all, because as the sellers ask the question, they get very confident answers from the AI, but seller confidence dropped like a rock as soon as they started to see that some of this was confident but a hallucination. Some of it was confident, but the context wasn't right for the situation they were in. And it was very confident, and reps were like, "I can't. I don't have time to fix this." And so there were all kinds of strikes against what was coming out of Copilot, and it wasn't Copilot's fault. It's not Claude's fault. It's not Chat's fault. It's doing exactly what it needs to do. It's not Salesforce Gong and Seismic's fault either, because they're doing the thing they're doing. They're just providing some signal. It's what happened in the middle when nobody could tell, the Copilot couldn't tell what good looks like. And it was getting so much signal that it couldn't judge or discern what the right thing was to then assess the situation for and provide recommendations for the next best actions. It was literally paralyzed into standard generic, albeit confident, Judge, uh, uh, advice, but not solid discernment and judgment on what good looks like and what will provide impact. So from here, I want to take everybody to like, well, what are you going to do about it? So this is the crossroads we're living with. Some people maybe have Copilot and, and they just let people loose on that. Some people have attached it to their proprietary LLM, and some have maybe done all these integrations I just shared. But in every case, I think people are less than satisfied with the outcomes they're getting. And before people give up on it, we're hoping to right that wrong. So let's sort of— I'll call this a maturity model, but it's not a maturity model. It's really thinking through the phases of how AI coaching has evolved. So think generic AI. This is where you get your AI device— Chad, Claude, Copilot. So this would be— Ah, I've given everybody Copilot. And, and then recognizing that it gives you only a so-so answer, not deeply personalized, not very context aware. The second is when you start doing integrations to all of the platforms and data and signals that I just shared. And context aware AI is that now at least you've applied some deal context, some previous calls to it. There's account history, there's contact information, there's a layer of information. That comes together with your AI to make it less generic and context— now contextual. But this is where the problem I just described happens. What, what we're looking for and what's needed for AI coaching to move forward is that you need a methodology to inform the AI. Like, something has to be the arbiter of what good looks like and can provide sort of a predictive assessment layer that that says, this is what the signal means, and then a predictive coaching layer that says, now here's what you should do about it, and in a way that is way more confident. It leverages everything below the line, right? This is all coming to bear, but now you have a methodology layer that is informing the AI, like it filters all of those signals through and to the Copilot to say, here's the better way to handle this. But we think there's even a future that's greater than that. And that is where the methodology is going to govern the AI, where you now are thinking about less, did you add the methodology to the soup and hope that the soup tastes better? Or are you going— I don't even know if I can stretch this metaphor, Gerhard. Or are you going to literally have the methodology run that, uh, I'll call it intervention. Um, so let me— this slide will help explain it. Methodology-governed looks more like this. Like, the methodology is looking at all the signal and diagnosing what the situation is. It can tell you what's going on because it knows what good looks like and it knows what it's looking for. Then it says what you need to do next is apply this framework because this is the situation. It's confident in the situation and correct because it's born out of a methodology that's been proven to work and now just needs to be applied consistently in a high-quality way at every step or stage and for every rep and every deal. The third thing it has to do is then like evaluate it against validated standards. This is what I mean. Like this is what good looks like. And this is the situation you're in. This is what you need to do in this situation because this returns the best results if you do this. And then keeps you within guardrails. Like a lot of people are like, I want to go here. I want to do this. I think I should do that. Sometimes you need to be coached and your manager might put guardrails on you, but they can't get to every step and every stage of every deal for every rep. So the methodology inside the AI is providing those guardrails for what you should do in a given situation, which means you can get to a recommended Predictive next action, meaning here's how you should call prep this. Here's what you should do in execution. And then when you're done, it listens to that call and it prescribes the follow-up. All you know daisy chaining these all together, coming down to this. And now I'm in a position to have my methodology activated for me in this stage for this deal and this rep.

So let me jump in because I want to clear up a question that how would you define methodology versus process?

This is, again, I didn't even send you my slides, but that's what I'm going to do next.

Like, okay, let me make a comment is that if you do not have a methodology, governing AI, you are going to suffer from the pathology of AI slop.

Yes, that's all that's really left. And the thing about it is, if you're not discerning or have good judgment, you can't tell it's slop because it is so plausible sounding sometimes and looking. It's just amazing. And I don't have time to talk about a new study that we're doing with Dr. Lefboni at Florida State. We'll have to come on another webcast. Where he is studying cognitive surrender when salespeople just use the slop and when others use discernment and watching the outcomes. And he's got thousands and thousands of instances now, and he can see that cognitive surrender is real and it results in poorer outcomes. So we do have to get better at judgment and discernment. And is that going to be up to the individual rep, or are you going to give them a governing layer, uh, that helps filter this and improve that outcome? So I love your question. Because is it process or methodology? The answer is both and because in my definition of methodology, you are integrating 3 things: the process or steps that you need to follow, the skills you need to execute and create a differentiated experience, and the story you tell to bring it all together. To me, the a big M methodology inside of oops inside of any company. Is is this idea that you are able to knit these together, right? That the the winning human interaction is you following the right steps, executing the right skills, and telling the right stories, which gets you from you the company to a customer saying yes. And that looks like this in practice: You have to have stages. You have to have steps. Let's call this process. You got to qualify well. You got to run a good engagement. You got to administer that thing right. You got to understand the verifiable outcomes, the buyer mindset, and the key activities to run each stage because you got to run a good engagement so a company can manage this, coach this, and can predict their pipeline and have some confidence. But run an engagement as well as you want. If you show up or don't show up in a way that you deliver that as an experience that sets you apart when everybody's running sort of similar stage-gated processes and everybody's selling somewhat similar stuff? Like, what's going to set you apart? That's where we say it's the skills of a salesperson with their lips moving that kind of brings that stage to life. So if I have to gain problem alignment and agreement, I need to be great at interacting in a way in my discovery that aligns solutions to needs. And delivers a problem statement that everybody agrees on so that we can move to here. So every stage actually has a corollary sort of competency or skill set to make it come to life. And sometimes people put the stages in place and they think that makes people better at that stage, when in reality there are associated behaviors and skills with each stage that have to be trained as well. And I say, and then messaging and story that That kind of correspond to where you are in this process and can help move this thing along. So your story, your skills, and your steps need to be in alignment to call it a methodology. And, um, and that's interesting because, um, uh, if you think about it, that's where playbooks have kind of come from. Like, we're going to put our steps in the playbook, and then we're going to say, and these are the skills you need for that stage. And here's the messaging and stories you need for that stage. And people have built these mammoth playbooks and have tried every which way to get sellers to use them. And, uh, they don't lay on shelves anymore. Now they just collect dust in the, in the digital, uh, shelf. Um, but like, it's, it's massive because it's not only hard if you think about it for one stage. Now you have to think about every stage needs to have the right steps, the right skills, and the right stories. For each stage, so now a seller is either like, "Can I memorize that, or am I going to go to the playbook each time?" What I'm saying is it's right in spirit, but it's not scalable. It's hard to execute methodology when all three layers are supposed to be interoperating. So that's where you now have to start thinking about the power of AI. The power of AI. Is that it's going to help run the methodology, right? Slice— well, I'll just build this out, like slice by slice. So imagine that you have an AI coaching assistant that is orchestrated, is for each stage, and it's orchestrated together to look at the signal, assess the situation or stage you're in, bring the right framework to bear for your process, Bring the right skills to bear to bring that life in a differentiated experience, and brings the right messaging and story to bear so that you can execute that stage. So what we're saying is your AI coaching needs to activate, and then it needs to be enriched, and it it needs to then be situationally fluent. And what that means is imagine your AI. Your copilot now is able to help you call prep and call follow-up each stage, integrating the 3 things on your left: the steps, the story, and the skills. I'm going to activate that so that you prep well for that call, you execute it well, and you do good follow-up. It gets even smarter and better when you can integrate your CRM, your conversation intelligence. So let's say for example's sake, your Salesforce, your Gong, and your Seismic, like my example at the beginning, because once you then can pass the activation through this enrichment is when you get, you get the, you get the gold. Like you can, you have a predictive judgment layer that assesses all of that and says, this is the coaching you need and tells the manager, this is where you can help. And AI coaching needs to have to get to this place right here in a way that's correct versus just confident. It needs to have the methodology activated and then integrated to that signal, and then the outcome is is is correct, not just confident. Now, will it look terribly different? Because It's still probably a text-based nudge with maybe some simple visuals. It's still going to come out looking like, well, that looks like the same stuff, but it's going to be wholly different because when you examine it, it's, it's going to be governed by the things that we talked about earlier. And now you can, you can move out with that coaching knowing that it's correlated to some level of cause and effect and, and you are working Correctly, not just confidently resharing what AI did for you.

So how do you create that governor for AI, for the Copilot?

So you— what we're doing, because we are a company that, if you don't know, those of you who have come here, we have built a sales training curriculum And really a sales methodology content IP library. So Corporate Visions is a company that has built sales process for winning new business and managing opportunities, built account management process for renewing and growing existing accounts. We've built skills and competencies for both, and we've built messaging frameworks for both. All of it backed in research. So any of those things have already had research applied, and what we've had to do, and what I think you need to consider, is like then how does that get built into what is technically being referred to as a knowledge graph? Markdown is the technical thing, but what we've done is building a modern markdown knowledge graph on all of. The sales process skills and stories knowledge that we have using an ontology that organizes these things so that it can provide judgment and governance. So I'm just going to show you one example here. Don't panic, anybody. On the previous page, you saw that we had identified all those competencies and all those stages. There's basically 17 total— we'll call them nodes. One of those, let's say, is negotiating competently and creatively. What you're looking at here is everything that went into the knowledge graph, all the research that we've done. So we do proprietary, original, exclusive research on each of those 17 areas. So we have research, there are frameworks, concepts, and models that are applied. So those are put into the knowledge graph. So call it concepts and frameworks and specific skills that underlie the competencies, right? And then there's what's the buyer state like? What are they thinking? What are we have all that because of the process work we've done. There's behaviors. These are the things you need to do and the things you should expect to see from the customer. This is the language you should expect to hear from the customer and and some other. Context. Oh, and there's always job aids to help you build out that particular competency, like have a response to the customer. So what you need to know is that for each one of 17 specific competencies and stages in new business acquisition, retention, and expansion, we have loaded up what we're calling the brain. With the related 51 skills, the 348 behaviors, and all of our research sources. And now it looks like essentially this. There's a brain with all of that intellectual property in there. And every time we have new research, it's updated and appropriately applied. And we have an assessment judgment layer around that because we now know what buyer experiences matter and what those look like. So that level 2 methodology informed, level 3 methodology governed. If you want to inform your AI, you can connect to this directly to this knowledge graph, and it helps inform and make your AI better at filtering and providing correct answers. If you want to go all the way and say we need to lock this down to govern every move. So that the AI is doing all the things I said, from assessing the situation, validating the frameworks and guardrails, and then presenting next best actions within the context of assessing the signal that it's getting, there's another layer of solution. And so, it's— what's happening is imagine that all the kinds of training that you could give and all the kinds of manager coaching you could give for every situation possible that aligns to a buyer experience that's predictable is now sitting in this knowledge graph or this brain, and it can be distributed, uh, across, uh, those experiences stage by stage, deal by deal, rep by rep, when they go to their Copilot, their Claude, or their ChatGPT, um, or name your, your AI of choice.

I have—

I'll pause there for a second.

I have a name for this.

Yeah, I need one because CVI brain isn't that great.

So, uh, it is AI cybernetics. Um, when you think about Norbert Wiener, um, who, um, gave it the term from the Greek kybernetics, it means steersman. So cybernetics focus is not on what a system is made of, but on how it behaves and communicates with itself and its environment. So essentially, it's a system to self-regulate, adapt, and achieve goals.

Huh. Well, I'm going to see if that URL is available for this application. But I think you're right. It's where we are, where the technology allows us like to be. But yeah, to the point, like the risk is there that you're gonna get confident answers that are not operating within a well-defined system that has some, again, proven cause and effect for business outcomes and impact. It's just a scrape of the web of plausible sounding blogs and books and everything else.

And I give you an example. You can go to, let's say, Claude. And feed it an article that you have written, let's say 300 words, and say Claude transformed that article into the style of David Brooks, the New York Times columnist. And it does it in 14 seconds. But then you give it the same command, a similar command, and say, how would Mel Brooks rewrite that paragraph? And you get something totally funny.

No. Well, I'll give you a quick story. My daughter's a BDR, and the other day she sent me a cadence she had produced in her sales engagement platform of voicemails and emails and call scripts. And she told me her prompt was, she went into AI and said, build me a multi-touch cadence of voicemails, emails, and call scripts based on Tim Reister's book, Conversations That Win, and his Why Change framework. And she got an answer and she sent it to me.

And I'm going to be—

I was like, shoot, you're out of the will. I mean, like my own daughter is stealing my IP through Claude. But, you know, it's from a book. The book's like now 11 years old. And so, we have like our job is now to keep the IP fresh. It's now what's the modern buyer? What's the modern selling situation? How are we able to help a seller better engage? So the key is keeping this IP researched and fresh and always evolving, understanding the buying and selling situation. But I was miffed that, um, she could get something plausible out of a chat, uh, without paying for it. Uh, but you're, you're an author. It happens. Um, I'm, I'm, I'm in the same boat as Beyoncé. That's all I'm saying, Gerhard.

Well, they're stealing your song.

Right, exactly. Right? So I want to land here is what we're now seeing is the ability to move from sad copilot to happy copilot by putting a methodology-governed layer of judgment and coaching that takes in the signals from above and makes orderly sense of them, governs them relative to understanding where the rep is at on that deal in that stage. And so when the question comes from the rep, they're now getting a higher degree of confidence that they are getting the help that they need, um, all along the way— call prep, call execution, and call follow-up. And so it's, it's, it's where I think Anybody who's got training curriculum and content methodology they put in place, that's where the playbook has to go next. The playbook, the static playbook, even an interactive playbook showing up in the CRM is not going to get you where this is going to get you because AI is here and people are going to be looking for some coaching and they want it instantaneously and they want it right and they're not going to peruse your playbooks. And so We think that all companies are going to have to figure out how to get their methodology in as the filter for and the output of what good looks like.

So let me shift the conversation. How do you get from good to best?

Yeah, I think that it— if I don't have— oh, well, I'll just say it. Like, if you think about, we had 2 flavors, methodology-informed, methodology-governed. Methodology-informed simply puts some methodology into the mix. So you're hoping that Copilot or your AI will, let's say, lean on the methodology that's now being pulled into its thought process, but you're still hoping that it will use that Right or correctly when it's pulling in all this other information. Methodology—that's good. Like now, at least there's methodology in there, and there's a better chance that you're going to get something not sloppy. But methodology governed is going to give you the repeatability, the consistency, the scale, and and hopefully the quality every time out because its ability to interrogate the situation and give you what's best. And so. If you have a large sales force and many of them aren't like strategic account managers, they're everything else, you're gonna go from better to best by moving from informing with methodology to governing with methodology.

And then there's this very important word, judgment. Yep. How do you not only develop that, but also Apply it consistently so there are no shortcuts?

I'll answer that question as people can look at the, the code here to go get, um, the paper I wrote on all this in case, uh, you, um, didn't take enough notes and, and, uh, want to share it. So you can go get that. So judgment. So there's— this is the challenge, right? Most methodologies have been developed by people who try to codify Their personal successful performance, and they said this: I'm Mr. Miller and Hyman. This is what I used to do. Or you know, I'm I'm Mr. Bosworth, and this is what I used to do. I love all of them. Don't get me wrong. But I think the the question is: Is the methodology? And again, I'm I'm I'm putting process and skills and messages together. Has that been researched? Has that been proven out in an objective, rigorous way to say that, um, these are the experiences you need to create, and now here's the way a seller has to operate to create that experience? And we know that if you create that experience, you will win more often than not. And when we look at our judgment, we have a database of win-loss and retention-churn research The largest in the industry. It's 150,000 B2B decisions analyzed from the buyer's perspective. So the buyer feedback is what we've got, and we've been able to isolate those predictive experiences that we've then said, well, okay, we know if you do these well, you win. So what does good look like? Well, what are the skills that make you good at delivering that experience? Well, that's the research in the skill development that's in the portfolio that we have. And so your judgment comes from, is there any basis in some sort of at least correlated, if not cause and effect environment that says these experiences end in these results and these skills contribute to creating those experiences? And it's done in a way that is objectively validated, not unexamined folklore.

Tim, how can people get in touch with you?

I, um, I'm at Corporate Visions, um, uh, corporatevisions.com. I'm also on LinkedIn. But if you want, and if you liked anything you just heard from me, uh, here's a code to get you free, um, white paper that I've written on, um, all of these, uh, on basically the journey to methodology-governed AI. And, uh, that will, uh, be something that I think you can take in your rooms and talk about. And see where, where what you've got on hand, both in terms of AI and what you have on hand as far as methodology, can be joined together for better outcomes.

Thank you, Tim, for a fantastic session.

Thanks for the interaction. Keeps it alive. Thanks, Gerhard.