Lightly edited transcript: names and terms corrected, filler, false starts and crosstalk trimmed.

Francis Wade (FW): Welcome, if you’re joining us. I’m Francis Wade, your host, and I’m joined today by Amie Devero. Amie, welcome.

Amie Devero (AD): Hey, good to see you, Francis. I’m having more technical difficulties. I’m not used to doing anything media-related on a phone, so I don’t really know what I’m doing here. I’m trying to adjust the volume so I can hear you better.

FW: No problem. I’m also having a bunch of technical issues. But let’s jump into our problem of the day. The title, as folks may have seen, is “Hard to Engage Staff on Vision and Strategy? Do It in Your Sleep with AI.”

AD: I think that might be a gross overstatement.

FW: It could be. But it’s a real problem, isn’t it? Coming out of a strategy session or a retreat or a project, a CEO or a chairman or an executive wants to communicate the strategy and the vision to their staff. They want to engage them and get things going, because it’s not enough just to have done the strategy work. Maybe they tried it in the past and thought it would go well because they were very charming, or lucky, or good-looking, and it went really badly. Now that they’ve been burned, they realize they’re not Tony Robbins combined with Les Brown combined with Zig Ziglar. They don’t have that magic when it comes to inspiring others. But they have to do it again because it’s part of their job, and they’re reluctant even to start thinking about it. So that’s part of the problem we’re looking to solve.

AD: Hang on, I need to adjust something. [Audio adjustment.] All right, I’m not going to waste time on that; I’ll just try to decipher what you’re saying, even though it’s coming through exceptionally muffled. I actually think that whole context is a little off. I don’t know that communicating strategy requires being inspiring. It’s necessary to inspire people with a vision, but not with a strategy. And I also don’t think communication tells the whole story. What I see a lot, in my own practice but also in case histories, and if you follow the stock market and listen to earnings calls, is that a new strategy is developed, and then the CEO and the executive team go on an evangelistic tour. They do a bunch of town halls, put out a bunch of PowerPoints, have meetings with their teams, build balanced scorecards and cascade KPIs. But it’s finite, like a campaign. Whether they spend a month doing it or six months doesn’t really matter. At a certain point it stops, and then what the entire employee body is left with is the metrics. They no longer have any reinforcement for the strategy.

And they never really had the strategy translated into their own job. They only ever had the big-picture version, and maybe some departmental version from their VP. But they never got to address the question: what does this mean for me, in my choices between doing X and doing Y? Nor is it an employee’s job to even ask that question. So they’re left with metrics and KPIs. Success in their job is dictated by fulfilling performance metrics that are hopefully connected to the strategy. Maybe they are, and maybe they aren’t. Maybe they are, but they contradict each other, or drive the wrong behavior. Who knows? Because the strategy is now a memory. It’s not present. So when I look at that, I don’t see a communication problem. It’s not that it wasn’t communicated. What I see is an activation problem: it’s not active for employees the way it may be very active for the CEO, and almost certainly is for the board of directors. And that lack of aliveness is where the strategy goes to die. That gap between communication and activation is where strategy dies.

FW: I agree, of course. I wouldn’t have you on the program otherwise.

AD: Well, I don’t know about that. We could have a great argument about things we disagree about.

FW: Right. But someone listening (and I see we have a listener, thank God) might ask what you’re talking about, because every CEO and every chairman sees this as a communication problem. I’ve been reading about the solutions people are pushing on LinkedIn and on podcasts, and they always make it sound easy. It’s just a matter of telling people. All you have to do is make it simple and communicate. The word “simple” is often attached to communication as the solution. So the CEO thinks, “I need to communicate better, or more. I need to do more town halls, or send out more reports, or get my managers to communicate more with their staff.” It’s a more-more-more phenomenon, because what they’ve done in the past hasn’t worked. As far as they can tell, their part is to keep talking, or to get someone else to keep talking, or to get people reading.

AD: And I think the really telling thing is that in everything you just said, you could substitute the word “talking” for “communicating.” And communication is not about talking. For talking you could substitute typing, emailing, measuring. None of those things are communicating. They’re a one-way delivery of information. And those LinkedIn posts that say it’s easy (just talk more, just do a PowerPoint, just dumb it down, just turn it into a single rule) are all just delivering information. None of it becomes active in day-to-day work unless you have an exceptional manager who can take the strategy and use it in the day-to-day management of their team, as they make decisions on an ongoing basis. Do we do this or that? Do we put resources into this and take them away from that? Do we close this deal even though we’re not going to deliver the thing we promised the client?

FW: [Audio difficulties.] I apologize. Could you just do a recap for me?

AD: Sure. A couple of things. One is that just saying more is not the same as activating the strategy in the mind of the employee. So no amount of doing more as a CEO delivers activation, because activation has to live inside the employee’s experience. Probably the one case in which that happens is if you have an extremely strategic manager who, in the day-to-day management of her team, actually brings the strategy to the decision-making table. “We’re trying to decide: should we put this clause in the contract even though we can’t deliver this, and possibly hurt our retention rates? Or should we over-communicate with the customer and reset expectations, even though we might lose the deal?” That’s where the rubber meets the road, in these moment-by-moment decisions about, essentially, do we screw the customer later or now? And then the manager says, “Let’s look at our strategy. How would our strategy decide that?” Maybe they go blank. Or maybe, because they have such a killer manager, they’re with her, because it’s been an ongoing subject of conversation. And it is the context inside which decision-making gets done, which is what activation is: strategy as context. That’s the piece that can’t be delivered with more one-directional blah-blah-blah. It just doesn’t do that.

FW: [Partly inaudible.] …the manager has a harder problem than you’re describing, because now you have people looking at you with blank faces if you haven’t communicated it well.

AD: Your new technology is much worse than your old technology. It sounds like you’re swallowing every word. I can make it out, but it’s taking a tremendous amount of energy, which I’ll put in because I’m your friend, but our listeners may not. All right, I’ll talk amongst myself. No, but I think that points to exactly the problem. If you have the sensational, amazingly strategic and gifted manager who can bring the strategy to the moment-by-moment life of the employee, then you’re lucky. But I don’t know that executing strategy ought to be a matter of luck. You want a way to do this that is effective, consistent and reliable, and that doesn’t depend on the good fortune of having a brilliant manager who has both functional skill and that skill. So in the absence of the one-in-a-million manager, the question we wanted to ask is not only whether there’s a way to do this at all, but whether it could be assisted by AI. That’s something both of us have been working on, together and independently, and I think the answer is yes. There are a lot of ways to crack it. But what does it look like for strategy to be activated at the level of the employee? And by that I mean everyone, from the receptionist to the janitor to the customer service rep to the person writing code.

FW: Let me test whether you can hear me now.

AD: Oh my God, so much better, Francis.

FW: No more Bluetooth. Okay. Activation, then, is someone acting on behalf of the strategy, doing the kinds of things you mentioned, where they’re making trade-offs and they know the nuances—

AD: Or they’re creating the nuances.

FW: They’re creating nuances every day. They’re seeing holes in the strategy and they’re communicating those holes. The problem people have been trying to solve with open strategy is that the people on the front lines know the day-to-day and the people who develop the strategy don’t. You can fix that if your employees are activated, because they’ll communicate the holes in the strategy upward to the executive team. So it fixes all of the problems you really want fixed.

AD: I think you’ve gone a step beyond what activation fixes, and let me say why. If strategy is active for employees, it becomes the context for decision-making, yes. That’s really the goal of what we’re talking about. But the next step you go to, this feedback mechanism, doesn’t happen automatically by virtue of that, because that’s a process problem. Just because employees can make decisions in the context of the strategy, or can discern principles that may not be explicit in the strategy but turn out to be implicit, doesn’t mean they can identify strategic problems. What they’ll identify are things that, for them, are complaints. And that’s fine, because complaints are often symptoms of strategic problems. But I don’t think it follows that activated employees provide feedback that goes up. They may, they may not. They may just complain. They may just fail at executing the strategy. But one thing’s for sure: step one is getting the strategy activated, so they act with the strategy as context. And you can’t do that by talking to them.

FW: Right. Let me put off explaining why I said the upward feedback would happen until we discuss our different approaches to solving this problem. So let’s pivot to how AI can be used to crack the activation problem. I know we’ve been coming at it from different angles, and collaborating. Would you like to go first and share what you’ve been working on?

AD: Sure. The timing is perfect, because I finally have a prototype I can share. I’ve been fighting with a vibe-coding platform for, I don’t know, a month and a half, two months, three months? It seems like forever. I apologize to my subscribers; I’ve literally not published a word, because this is all I’ve been doing. Vibe coding is a very tedious process. Something as simple as rendering a matrix can take three weeks to tweak. And the AI iterates fast, but not fast enough for me to get into a flow, and too fast for me to go write a paragraph.

What I’ve been creating is a way for an organization’s strategy owner, whoever that is, to give an AI enough information that it can understand the strategy. Through an entire dialogue, an interactive process with the CEO or strategy owner, it maps the strategy against a competitive matrix and discerns principles, and comes up with a version of the strategy that the CEO says, “Yes, that’s it. You got it.” It doesn’t require any particular documents or any particular format; the process walks you through with whatever you have or don’t have. The CEO can then have the rest of the leadership team vet it and come back and say, “No, actually, this is a little off, and that’s a little off.” When they’ve got it all wrapped up and they have the strategy, they can disseminate it to all the employees. The employees go through an interactive onboarding to the strategy and can then use the AI to make real decisions.

The onboarding is again an interactive process, where the employees learn the strategy and then answer questions or give solutions to dilemmas based on their reading of the strategy, and the AI helps them internalize it and get the nuances. Then, as they have day-to-day dilemmas, if their manager isn’t the one they talk to, they can go to the AI and say, “Look, I’ve got this problem. This contract is about to close. The customer is expecting a feature the team won’t really have until next year, even though they’re saying there’ll be a beta in six months. Do I run the risk of closing the deal?” The AI can take the principles that have been discerned from the strategy, and the scenarios the CEO already worked through and answered, and help the employee make a decision that’s consistent with the strategy. So that answers part one, which is activating the strategy. And use of this tool can become part of the performance review process, so you can actually make sure employees use it. Otherwise they won’t, because it’s one more thing to do.

FW: It’s not optional. Especially if you’re in a strategically sensitive or pivotal role, a role the strategy relies on, where you need to do something different and the strategy needs to help you do that different thing. Because there are some people, like the gardener, or people who do customer service, who may not be engaged directly in the new strategy. They may just be doing business as usual.

AD: Right, like maybe the accounting team doing maintenance work that isn’t strategic. But anybody who interfaces with customers, or with service, or with product creation or product delivery (unless the strategy is some back-office financial strategy, and there are such things) matters, because most strategies rely on a particular positioning with respect to customers and product. That means everybody who touches those areas has strategic importance. There’s a big difference between how you sell a deal if your goal is retention and long-term customer trust, versus lots and lots of sales without caring about churn because you’re trying to show revenue growth. Those are different strategies, whatever our opinion of them might be.

FW: And the availability of the tool means that, literally, you could be sleeping, maybe taking a nap in the middle of the day, and your employees are accessing it—

AD: Or you could be in a different time zone, on the other side of the world, and your employees are using the AI to parse decisions and think through what is truly a dilemma, where both options are advisable and both are ill-advised. How do you choose between them?

FW: Right. We need to do a whole podcast on just what you’re working on. Coming soon. So let me share a little of what I’ve been doing with the time we have left. I’ve been focusing on the first part of the problem, which is the initial communication to staff, kind of like the onboarding. You mentioned before that the options have been pretty poor. You have a town hall and usually get no response. You send out a document that nobody reads because it’s too long. You rely on your middle managers, who themselves don’t really understand what they’re communicating, and who pick off some small piece to focus on, because the parts outside their understanding are too mysterious. So I’ve been focusing on how you introduce a strategy to people who weren’t there when it was developed. The techniques I’ve been using with a couple of clients are actually pretty simple.

AD: I think it’s very cool.

FW: The first part covers where the company is in terms of its past and its present. What’s the background? What brings us to today? What are the salient facts, the known challenges, the scary trends? It packages them into one very simple display. The second part looks to the future: what’s the target year (in my case usually 15 to 30 years out), and what does that year look like in terms of metrics, in terms of experiences, or other ways you could describe it? That’s a standalone part two. And the third part uses something like NotebookLM. The strategy documents could be a traditional report, but nowadays they can also include transcripts, videos, audio, survey results and interview results. If you want to get fancy, you can pull in industry material, though I think that’s less important. All of the materials that went into developing the strategy, if you capture them, can become sources for a NotebookLM notebook. If you’re familiar with NotebookLM, it lets you enter a closed environment and query its contents using natural language—

AD: You made that more obtuse than it needed to be. NotebookLM is an AI that lets you upload a bunch of sources. They can be links, documents, books, anything. You put up as many as you want; that’s your source list. Then you can ask questions about any or all of them. You can ask it to tell you more about something, or explain something, or ask what inference it would make based on any one or all of the sources. It’s an interactive interrogation of source material.

FW: Right. And you can ask it to cross-reference, compare and contrast. You can say, “What are all the mentions of this phrase in all the conversations we had to develop this strategy?” And it will faithfully answer. Someone said it’s a bit like getting a junior assistant to read everything and remember everything. They won’t be smart about the contents, but they’ll remember everything and faithfully tell you what’s in it, in a way human beings really can’t or don’t.

AD: And even if we could, we’re not going to. The other thing it’ll do: if you really don’t like the charts and graphs and introductory materials Francis gives you to explain the strategy, you can have it make a podcast of all those sources and listen to it. NotebookLM’s claim to fame is that it will turn anything into a podcast with AI hosts, so Francis and I don’t have to do it for you, which is fine with me because I hate being on camera. There’s also a new video feature, a video summary of some kind. I haven’t tried it yet, so I can’t tell you how well it works.

FW: Oh, I’m trying that today.

AD: So you’re thinking, what, AI avatars?

FW: No, no. It does a voiceover of a PowerPoint-style presentation; it puts together slides. Not that exciting. Avatars would be exciting. Not yet. Anyway, the point is that someone who wasn’t at the retreat, any staff member, could spend ten minutes or ten hours interrogating the heck out of the strategy. In other words: What does this mean for my job? Is there any application for this? Was this considered? Was there any thought that this new technology would have an impact? They can ask all of these questions in excruciating detail. And here’s why I said what I did about feedback. Ideally their questions, if they’re pertinent (and a bunch of them probably will be), are going to be about the assumptions that were made when the strategy was developed.

I’m working with a client now that’s actually rolling out this system. The problem is that the strategy was done two years ago, and I don’t think we mentioned AI other than in passing. I was facilitating, and I can’t remember the term coming up even once. I could be wrong, but it was not a big deal. Today it’s a huge deal for this organization. So they actually need the employees who are most familiar with AI, who on average are going to be the younger employees, to say, “The assumptions that went into that strategy two years ago are no longer valid, because of the new technology that’s emerged, and it needs to totally reshape things.” So there needs to be this upward, or at least this rejiggering of the strategy, because a new, disruptive technology is now available. If you allow people to engage, they will have something to say, and it might be important.

AD: To that point, I just wanted to distinguish those two, not because they aren’t both incredibly important and ideal, but because I don’t think one necessarily produces the other at a process level. I will say that the tool I’m building (and for anyone who’s interested, and Anne, thank you for being here; you are currently both of our favorite person) is called Contextuum, and it will actually provide that feedback. It anonymizes the content of every conversation an employee has with the AI, but it infers implications from trends. Say multiple salespeople are complaining that features weren’t released on time and they’re having trouble renewing clients, and they go to the AI to ask how to renew in the face of not having delivered. If that happens more than a few times, it becomes a point of interest that the CEO or strategy owner eventually gets back on a dashboard. The dashboard also lets the CEO see recurring issues, recurring principles that are being questioned, points of contention, and frequent types of complaints. Because the AI anonymizes all of it, it provides that feedback in a way that really cloaks the identity of the people involved, while providing important iterative insights that can inform tweaks to the strategy, rethinking of assumptions, or even rethinking of entire strategic themes. Why build a tool that activates strategy if it doesn’t provide an ongoing feedback mechanism? Strategy is dynamic. Even in the tool I built, every piece of the strategy can be revised at any time by the people with permission to do so.

FW: Right. The thing we discovered, folks, all in the last three weeks—

AD: A month, maybe.

FW: Maybe a month. We couldn’t have done this six weeks ago, because we hadn’t even had the initial conversation. What’s different about what we’re doing, compared with what I’m hearing about elsewhere, is that this involves people. It’s driven by stakeholders. Here we’ve talked about staff, but it could be any stakeholder. It’s not driven by AI searching out the corners of the internet to find where your strategy may have gone stale. That has its place. But this is about activating every single stakeholder (employees, board members, you could even get customers involved) in a systematic way, using a dashboard and a language model. It’s people using AI, not AI instead of people. And I’m not hearing anybody else describe what we’re talking about.

AD: What’s been so interesting as I’ve been building this, because I’m not an engineer, is that I’m actually learning a lot about coding because of all the bugs that keep getting created. And I’m the only thought partner the AI has. Half the time I get annoyed with it: “No, stop doing that same thing. Let’s try it this way.” And the AI says, “Oh yes, we could do that.”

FW: It’s very obedient.

AD: Holy cow. Really simple things, like “Stop forcing it into a container that’s all lateral. Wrap the text.” Three weeks on a matrix rendering. But I digress. What’s been really interesting is the few things I’ve asked the AI to do within the program. For instance, I ask it to go out and find competitors: take the initial document the CEO has given you, and now populate some competitors. It’s terrible at it. It can find the competitors, but since every competitor, including the user’s organization, is rated on competitive axes like expensive versus budget, the AI rates everything dead center. It can make no determinations; it has no distinctions at all. So I stopped having it rate anything. It populates competitors, but everything has to be completely editable, so the user can delete all the competitors if they’re irrelevant and add their own. Almost everything I’ve had the AI populate through its own research has had to be either dispensed with or made entirely deletable and editable. But everything I’ve asked it to do in terms of discerning, inferring and extracting principles from the truth it’s been given, it does expertly. And that’s the piece you can’t replace without humans, except humans can’t do it as fast, or as consistently, for every employee, in every instance, across all the myriad use cases.

FW: Right. We need to wrap up. Thank you for showing up. We have four people, plus me. And Amie, thanks for jumping on at short notice; we talked this morning and here you are. Folks, you can follow both of us on Substack, and if you think we should do another one like this, let us know.

AD: And we’ll make sure to let you know when we’re doing it and give you some notice, since we only decided an hour before. I think we can schedule it.

FW: I don’t know if we can schedule it here, but we can schedule it and then tell our subscribers when it is.

AD: Right. Manually.

FW: Great. Folks, thanks for joining us. Take care.

AD: Thanks, Francis. Bye.