On this week’s episode of Inside Outside Innovation, we talk about why AI agents should speak Latin, how to avoid the agentic convergence trap, and the power of the peanut butter and jelly sandwich. Let’s get started.
Inside Outside Innovation is the podcast to help innovation leaders navigate what’s next. Each week, we’ll give you a front row seat into what it takes to grow and thrive in a world of hyper-uncertainty and accelerating change. Join me, Brian Ardinger, and Mile Zero’s Robyn Bolton as we discuss the latest tools, tactics, and trends for creating innovations with impact. Let’s get started
Podcast Transcript with Brian Ardinger and Robyn Bolton
[00:00:30] Brian Ardinger: Welcome to another episode of Inside Outside Innovation. I’m your host, Brian Ardinger, and with me I have Robyn Bolton. Robyn, hello, how are you today?
[00:00:47] Robyn Bolton: I am great today. How are you?
[00:00:49] Brian Ardinger: I am doing well. We are coming off our all directors meeting this week for Nelnet, where we brought in about 250 of our directors from all over the world to talk about AI and other things that are going on in the business, and it’s been quite eye-opening. So that’s where I’ve been. I hear you’re out on a beach.
[00:01:05] Robyn Bolton: Yeah, I’m kind of doing the opposite of your highly productive week. I’m being highly unproductive and am on vacation here in Turks and Caicos. So, I just dragged myself in from laying poolside so that we could talk about innovation, because there’s some really, really cool stuff going on.
Why AI Agents Speak Latin
[00:01:21] Brian Ardinger: We’ll get you back to the beach very shortly, but I always appreciate your insights. We will jump right in. The first article on the list is from Tristan Kromer. In his blog, he talks about why my agents speak Latin. I thought this was a very interesting take by Tristan. Tristan, if you’ve been following the podcast, he actually came and spoke at the IO Summit most recently.
He was talking about, in this blog post, about how he is doing a lot more coding and creating AI agents, and he has kind of a unique way to see if the agent is going off the rails. So, what he’s done is he creates these personas for his agents, and one of them he calls Cicero, and has it speak to him in Latin phrase to give it a personality.
And what he’s found, by giving his agents particular personalities and quirks, that they actually come back, and over time we do know that AIs tend to lose the context. You know, they tend to drift over multiple rounds of dialogue. He says by incorporating this little tick of a personality into his agent, he can tell when the agent is drifting to the normal way because it stops speaking in Latin to him, or it’s, it drops phrases, or does a couple of other things that make it known that the context is getting too full and the agent’s going off the rails.
What I found so fascinating about that, not only the fact that it actually goes off the rails, but the fact that it was a way to give the layperson a way to understand if the AI was not doing what he had originally programmed it for. I just thought it was a great insight into it.
Using AI Personas to Detect Context Drift
[00:02:47] Robyn Bolton: Fascinating. It’s funny, because Tristan and I were actually talking about this at the IO conference over breakfast the morning of the conference, and he mentioned that his chief of staff spoke Latin. And we didn’t have a chance to get into why that was, so it was great to read this article, and it was also incredibly timely ’cause a couple of weeks ago I had decided to make the jump into using Claude Code and VS Studio and trying to get into the vibe coding.
And I told Claude Chat, “Hey, you are an expert in blah, blah, blah,” and pretty quickly I noticed the chat lost the thread. And was just being chat and not being the expert and the coach I had asked for, and which then led me to typing in all caps as I usually, is usually how our chats end up. And it was just interesting to read in this article the research about why that was happening, what it is kind of in the background that causes any of the chats to lose that thread and then this kind of little safety net, we put it in there. So great, super actionable article, and now I’m thinking of all the personas my little agents can have.
[00:04:00] Brian Ardinger: He was saying that, you know, when the tick disappears from the output, you know, the persona has left the building as well. So you’d kind of treat your quirk as your instrumentation panel and your dashboard. Quite clever.
[00:04:11] Robyn Bolton: Yes, very clever.
Beware the Agentic Convergence Trap
[00:04:13] Brian Ardinger: All right, the second article I want to talk about today is from HBR, and it’s talking about beware the agentic convergence trap. And in this case, they talk about three case studies that have recently, in 2024, a federal class action named six major hotel chains, Hilton, and Hyatt, and Marriott, and the gang, of alleging that they had shared AI pricing platform that had produced coordinated room rates across competing properties.
The hotels had never communicated directly, but the DOJ and the FTC called it price coordination. A second example of this kind of convergence is a regional grocery store chain replaced its human promotions planners with an AI system and trained it to look at all the same public market signals every competitor uses, and within two quarters of this promotional calendar had converged with the market leaders, and shoppers could no longer articulate why they preferred one chain over the other.
And then a third example, they talked about a national landlord had adopted an AI rent optimization platform and used by competing property managers, and it raised the rents in step with rivals it had never spoken to before. And so the US Department of Justice was looking into that as well. What I found fascinating, obviously, about the individual case studies, but the fact that relying heavily on AI as your determination of your strategy has some additional potential downsides if AI is converging the thought and converging the output across particular industries.
Why Human Judgment Creates Strategic Diversity
[00:05:38] Robyn Bolton: Two big thoughts as I was reading through this. One was I just loved this quote from the article. You know, it’s kind of asking why. Why is this convergence happening, and why haven’t we seen this before? Because everyone’s using all the same data. And, you know, the quote, “Human judgment introduces natural variation. One manager overrides the suggested action. One team moves slowly. One regional leader reads the market differently. Those frictions, often regarded as inefficiencies, are the mechanism that produces strategic diversity.”
And I think this, as we’ve talked about in past episodes, is why we need human in the loop, why humans are still valuable, and it made me think of so many companies I’ve heard that are using AI to brainstorm new ideas.
And, you know, the caution is, like, if you want radical innovation, if you want something radically new, AI isn’t going to help you. But this is also a caution for you’re going to get the same ideas as everybody else because you’re sourcing from the same kind of data set, and it’s the humans and the creativity of the humans and the different perspectives of the humans that not only create strategic diversity, but also when we’re brainstorming and innovating, create the same diversity of ideas that we need as well.
[00:06:57] Brian Ardinger: You know, back in the day, there was often people talking about, well, all you have to do is switch your consulting firm and get a different answer. And sometimes I, I… There were some convergences there as well. It’s like, oh- Oh, yeah … clearly you were thinking about the, you used consultants to come up with this exact same answer that this McKinsey gave to your competitor six months earlier. So, it’s-
[00:07:12] Robyn Bolton: I was going to say, but McKinsey does- … you know, design slides vertically, but BCD does it horizontally-
[00:07:17] Brian Ardinger: There we go. Okay … and that just..
[00:07:17] Robyn Bolton: makes all the difference.
The Organizational Immune System and Innovation
[00:07:19] Brian Ardinger: Well, it might work in the AI world, so we’ll see. They may have some life left in them. The third article I want to talk about is Elliot Parker and his Alloy Partners blog talks about the organizational immune system always wins.
And we’ve talked about this on a number of different occasions, but this particular article really lays out the fact that traditional organizations, you know, existing organizations that have business models that are working and such, have a very, very difficult, if not impossible, time to disrupt themselves or do things new or launch new products, because the antibodies within the organization are always processing to protect that particular organization, even if folks know that that is going to happen, and even if they have desire to not make that happen, the nature of the beast basically inhibits companies from innovating beyond protecting their core.
[00:08:07] Robyn Bolton: Organizational antibodies are real, and they serve the same purpose as biological antibodies. They’re there to keep the organization safe. And the more and more your function, your role is about keeping the organization safe, thinking, I always pick on legal, but legal, finance, accounting, you start getting into supply chain, the more and more it’s your role is about driving risk out of the system.
Innovation is inherently risky, and so there will always be not just pushback, but talk about swarming and killing. It’s what antibodies do, and they will swarm and kill your idea, swarm and eliminate you if they have to. And so it’s something that you really have to be on guard with. And again, it kind of goes back to why innovation really is a leadership problem, because leaders are the best defense against those antibodies.
Why Innovation Often Needs to Happen Outside the Core
[00:09:05] Brian Ardinger: And you know, Elliot talks a lot, obviously he’s got a vested interest in this particular point, but the idea that the only way you can actually do this is to go outside and create entities outside of the beast to correct the architectural answer, to validate whatever the new is and move faster than you can within the given walls.
And there are very few companies that do that well as well. So, it’ll be interesting to see how the world evolves, how AI is going to affect all this. If, you know, new startups can use tools that give it some of the same advantages of a bigger organization, will they be able to move faster and avoid some of these antibodies? We’ll find out, I suppose.
[00:09:44] Robyn Bolton: Yeah. We will find out.
Uncrustables, PB&J, and Consumer Product Innovation
[00:09:47] Brian Ardinger: All right, the last article, a little fun one. I was talking to a couple people on my team who are addicted to Uncrustables, and so there’s a new article from Consumer Girl out talking about PB&J is no longer just a sandwich, it’s a growth strategy. This article goes into the, the world of the Uncrustable and the amazing growth.
And if you’re not familiar with it, this is Smucker’s Uncrustables, the little peanut butter and jelly sandwich in a little pocket. It’s growing at 20% year over year. It’s planning to hit a billion dollars in net sales by the end of fiscal year 2026, and the company just opened a 900,000 square foot manufacturing plant in Alabama to meet the consumer demand.
In addition, you’re seeing all these new competitors, private label dupes that have started to appear in the Costco shelves and such. I just thought it was interesting topic to talk about. Again, we’re talking about innovation, and even in an industry where peanut butter and jelly sandwich has been around quite a bit of time, how new innovations or new ways of thinking can actually change the trajectory of even the peanut butter and jelly sandwich.
How Small Product Changes Create Big Market Growth
[00:10:47] Robyn Bolton: Yeah, and you can never go wrong talking about or eating PB&Js. This is perfect. I did love all the other innovations within and beyond Uncrustables. You know, they talk about that there’s a brand that has launched a better-for-you Uncrustable PB&J that Smucker’s is putting out a high protein Uncrustable.
Gotta jump on that protein bandwagon. That there’s, like you said, there’s all sorts of other different versions coming out, and, you know, it’s just really fun. Even getting into the idea of Reese’s PB&J and then peanut butter powder, which I don’t know, I’m going to… I am a lifelong Jif creamy peanut butter. I will not eat anything else, so I’m going to reserve judgment on the PB&J powder, but I’ll let someone else go there first.
[00:11:39] Brian Ardinger: I will stick with Jif, but I’m a crunchy guy, so
[00:11:42] Robyn Bolton: Ah.
[00:11:43] Brian Ardinger: It’s interesting to see how these particular trends come into play and then what happens in the marketplace when these trends do take off. Like, they said it took a, the innovation from Smucker’s to come up with this pre-made sandwich, basically.
Because of the growth, now you’re seeing all these other folks that are both directly imitating it as well as taking the theme of the new product and putting it into spaces and that. Always fun to see how new things are created and what resonates with the market itself.
[00:12:07] Robyn Bolton: Yep, and I bet that was based on insights of lots and lots of moms spending lots and lots of years cutting the crust off of their PB&Js.
[00:12:15] Brian Ardinger: Excellent. That wraps up another episode of Inside Outside Innovation. Thanks for coming out. We’ll see you next time.
[00:12:22] Robyn Bolton: See you next time.
[00:12:26] Brian Ardinger: That’s it for another episode of Inside Outside Innovation. Today’s episode was produced and engineered by Susan Stibal. If you want to learn more about our teams, our content, our services, check out insideoutside.io or if you want to connect with Robyn Bolton, go to MileZero.io, and until next time, go out and innovate.
Articles Discussed
- Why My Agents Speak Latin | Tristan Kromer | Kromatic
- Beware the Agentic Convergence Trap | Patrick van Esch, Yuanyuan Gina Cui, and J. Stewart Black | Harvard Business Review
- The Organizational Immune System Always Wins | Elliott Parker | Alloy Partners
- PB&J is no longer just a sandwich. It’s a growth strategy | Sanjana Pattanaik | consumer girl
